INFORMAL PUBLIC HEARINGS FOR THE PROPOSED RULE

ON OCCUPATIONAL EXPOSURE TO

RESPIRABLE CRYSTALLINE SILICA

+ + +

UNITED STATES DEPARTMENT OF LABOR

OCCUPATIONAL SAFETY & HEALTH ADMINISTRATION 

+ + +

March 19, 2014

9:30 a.m.

Frances Perkins Building Auditorium

200 Constitution Avenue, N.W.

Washington, D.C. 20210

	

BEFORE: 	DANIEL F. SOLOMON

	   	Administrative Law Judge

U.S. DEPARTMENT OF LABOR (DOL):

KRISTEN LINDBERG

Attorney, Office of the Solicitor

ANNE RYDER

Attorney, Office of the Solicitor

IAN MOAR

Counsel for Health Standards, Office of the Solicitor

OCCUPATIONAL SAFETY & HEALTH ADMINISTRATION (OSHA):

WILLIAM PERRY

Acting Director, Directorate of Standards and Guidance

STEPHEN SCHAYER

Office of Physical Hazards

TIFFANY DeFOE

Office of Chemical Hazards - Metals

JANET CARTER

Office of Chemical Hazards - Metals

B.J. ALBRECHT

Chemist, Salt Lake Technical Center

JOSEPH COBLE, Sc.D., CIH

Director, Office of Technological Feasibility

DAVID O'CONNOR

Director, Office of Chemical Hazards - Non-Metals

DANIEL JOHANSEN

Chemist, Salt Lake Technical Center

PEER REVIEWERS:

KENNY CRUMP, Ph.D.

GARY GINSBERG, Ph.D.

BRIAN MILLER, Ph.D.

Institute of Occupational Medicine

 

U.S. CHAMBER OF COMMERCE:

HENRY CHAJET

Attorney, Jackson Lewis

JONATHAN BORAK, M.D.

President, Jonathan Borak and Company				

CHRISTOPHER M. LONG, Ph.D.

Principal Scientist, Gradient 

PETER A. VALBERG, Ph.D.

Principal Scientist, Gradient 	

PATRICK HESSEL, Ph.D.

Epidemiologist

THOMAS A. HALL, Ph.D., CIH

Former Professor, University of Oklahoma

ROBERT LIECKFIELD, CIH

Bureau Veritas HSE Laboratory

GERHARD KNUTSON, Ph.D., CIH

President, Knutson Ventilation

WILLIAM B. BUNN, III, M.D., J.D., M.P.H.

Consultant, Navistar

AMERICAN CHEMISTRY COUNCIL (ACC):

NEIL KING

Attorney, American Chemistry Council's Crystalline Silica Panel 

LOUIS ANTHONY (TONY) COX, JR., Ph.D.

President, Cox Associates

OTHER PARTICIPANTS:

DAVID F. GOLDSMITH, M.S.P.H., Ph.D.

George Washington University

DR. ROBERT PARK

Risk Evaluation Branch, NIOSH

CHRIS TRAHAN

Building and Construction Trades Department, AFL-CIO

FRANK HEARL

Chief of Staff, NIOSH

ROSA KEY-SCHWARTZ

Division of Applied Research and Technology, NIOSH

DARIUS D. SIVIN, Ph.D.

Health and Safety Department, United Auto Workers

INDEX

										PAGE

		

INTRODUCTION

Judge Daniel F. Solomon					  	   PAGEREF Intro \h  267 

INTRODUCTION OF PEER REVIEWERS

William Perry								   PAGEREF IntroPeerReviewers \h  267 

U.S. CHAMBER OF COMMERCE 

	Jonathan Borak, M.D.		 			   PAGEREF Chamber1Borak \h  272 

							 	

	Christopher M. Long, Ph.D.	 			   PAGEREF Chamber2Long \h  297 

 

	Peter A. Valberg, Ph.D.	 				   PAGEREF Chamber3Valberg \h  309 				 

	Patrick Hessel, Ph.D.		 			   PAGEREF Chamber4Hessel \h  323 

	Questions								   PAGEREF Chamber5QbyPeers \h  337 

AMERICAN CHEMISTRY COUNCIL (ACC)

 

	Louis Anthony (Tony) Cox, Jr., Ph.D.		   PAGEREF Chem1Cox \h  375 

 

	Questions 							   PAGEREF Chem2QbyPeers \h  400 

U.S. CHAMBER OF COMMERCE 

	Thomas A. Hall, Ph.D., CIH	 			   PAGEREF ChamberTwo1byHall \h  432 		

	Robert Lieckfield, CIH		 			   PAGEREF ChamberTwo2byLieckfield \h  441 

		

	Gerhard Knutson, Ph.D., CIH	 			   PAGEREF ChamberTwo3byKnutson \h  447
	

	William B. Bunn, III, M.D., J.D., M.P.H.	   PAGEREF ChamberTwo4byBunn
\h  461 	 	

	Questions 							   PAGEREF ChamberTwo5byPeers \h  471 

ADJOURNMENT								   PAGEREF ADJOURN \h  549 

EXHIBITS

EXHIBITS		DESCRIPTION				 	PAGE 

Exhibit 6		Reserved - Chamber of Commerce 					PowerPoint     				  
PAGEREF e6m \h  336 	     

Exhibit 7		American Chemistry Council 

			PowerPoint 			  		   PAGEREF e7mr \h  429 	

Exhibit 8		Silica vial, 5.6				   PAGEREF e8_10mr \h  470 

Exhibit 9		Silica vial, 12.7				   PAGEREF e8_10mr \h  470 

Exhibit 10	Silica vial, 25.5		  		   PAGEREF e8_10mr \h  470    

Exhibit 11	Reserved - Photo of silica vials	   PAGEREF e11mr \h  470 

Exhibit 12	Checkoway study			  	   PAGEREF e12mr \h  496 

	

P R O C E E D I N G S

(9:30 a.m.)

		JUDGE SOLOMON:  Okay, we're on the record. 

		I'm Daniel Solomon.  I'm an administrative law judge with the United
States Department of Labor.  My address is 800 K Street, Northwest, 4th
Floor, Washington, D.C.  The zip code is 20001-8002.

		To get started this morning, Mr. William Perry will introduce the
peer reviewers.  Is that right, Mr. Perry?

		MR. PERRY:  Yes, Your Honor.

		JUDGE SOLOMON:  Just state your name again, for the record, because
some of the people weren't here yesterday.

		MR. PERRY:  Thank you, Your Honor.  

		I'm Bill Perry.  I'm Acting Director in the Directorate of Standards
and Guidance for OSHA.  Today and tomorrow, we have most of the members
of the expert panel that conducted the peer review on OSHA's preliminary
health effects analysis and quantitative risk assessment.  They are here
in the audience to listen to witnesses today and tomorrow.

		We have Mr. Bruce Allen, Dr. Kenny Crump, Dr. Gary Ginsberg,
Dr. Brian Miller -- all the way from Scotland, and Dr. Andrew Salmon.
 And unfortunately, two of our other peer reviewers, Dr. Murray
Finkelstein and Dr. Noah Sayshus (ph.) did not -- had schedules that
did not permit them to attend, but we do have most of the panel present
here today.  Thank you.

		JUDGE SOLOMON:  Okay.  So the presentation this morning is from the
United States Chamber of Commerce.  Do you want to introduce the members
of the panel?  First of all, would you state your name for the record,
please?

		MR. CHAJET:  Good morning, Your Honor, members of the panel, peer
reviewers, and members of the audience.  Thank you for being here with
us this morning.  I am Henry Chajet.  I am a shareholder at the law firm
of Jackson Lewis.  We are of counsel to the United States Chamber of
Commerce.  We have a series of scientific, engineering, and economic
presentations that are written and in the rulemaking record.

		We will present summaries of those today, this afternoon, and we are
on for another day, I believe, on Friday.  So with that, let me
introduce my witnesses, if I may.

		JUDGE SOLOMON:  Sure.

		MR. CHAJET:  First of all, I just want to say a word about the Chamber
of Commerce.  The Chamber is the largest business organization in the
world.  It represents the interests of more than three million
businesses of all sizes, sectors, and regions, from small to
multinational corporations, and local chambers to leading industry
associations.

		To address the OSHA proposal, the Chamber has engaged a panel of some
of the nation's most distinguished scientists, physicians, engineers,
and experts on occupational health, as well as leading experts on
regulatory economics and on statistical analysis.

		Today we present some of that testimony focused on the science, and
later we'll present on other issues.  So our expert witnesses are led
and chaired by Dr. Jonathan Borak, who is behind me and to my left. 
Dr. Borak is a clinical professor of epidemiology and public health, a
clinical professor of medicine, and a member of the Yale University
program in occupational environmental medicine.  He's also President of
Jonathan Borak and Company, a professional services consulting firm with
extensive expertise in occupational and environmental health.

		Also on our panel and sitting to my right is Dr. Peter Valberg. 
Dr. Valberg is a principal at Gradient, an environmental consulting
company in Cambridge.  He works in the areas of human health risk
assessment and inhalation toxicology.  He has 30 years of experience on
the faculty of the Harvard School of Public Health, where his research
centered on how inhaled particles and gases affect cell function, and
how particle dose relates to the risk of lung disease.

		At Gradient, Dr. Valberg has provided air quality and public health
expertise to the Department of Justice, the United States Environmental
Protection Agency, the National Academy of Sciences, and a variety of
private and public clients.  He's a fellow of the Academy of
Toxicological Sciences, and he's authored more than 100 peer-reviewed
articles on how environmental exposures might lead to adverse health
affects.

		Dr. Chris Long is another one of our experts on the reviewing panel. 
He is a principal scientist at Gradient, and he specializes in air
pollution exposure assessment and inhalation risk assessment.  Prior to
joining Gradient he received his doctorate in environmental health from
the Harvard School of Public Health, where he conducted a research study
to characterize particulate matter mass concentrations, size
distributions, and chemical composition inside and outside residential
homes.

		Dr. Long has prepared approximately 25 peer-reviewed journal articles
or book chapters in the general areas of indoor and outdoor air
pollution exposures and risk assessment and on inhalation toxicology. 
He is certified as a diplomate of the American Board of Toxicology.

		Also joining our panel is Dr. Patrick Hessel, sitting behind me and
to my right.  Dr. Hessel is an epidemiologist with a focus on
occupational environmental lung disease.  He holds a master's in
environmental health from the University of Minnesota, and a doctorate
in epidemiology from the University of Pennsylvania.  He has conducted
research on pneumoconiosis, specifically silicosis and asbestosis,
asthma and respiratory cancers.  I will introduce the rest of our panel
in the next part.

		JUDGE SOLOMON:  Okay.  We're sort of pressed for time this morning,
and the Chamber of Commerce has advised me that they're going to take
about an hour and a half.  We will have some questioning from the peer
reviewers, then we will have some questioning from the public, and then
OSHA will comment.

		Now, we'd like to be able to do all of that before lunch.  We have
another panel that's scheduled to come in at 1 o'clock.  So do you want
to begin with your presentation?

		MR. CHAJET:  Yes.  Dr. Borak will chair our session and begin with
his presentation.

		DR. BORAK:  Thank you.  Good morning.  My first question, actually,
before I do anything, is to make sure I've got the right buttons.

		I am summarizing comments that I've submitted, and I've expanded them
a little bit.  And what I would like to do is provide a background which
really addresses a single question, which is the extent to which there
are data that reflect on the adequacy of the permissible exposure limit
of 100.  And my premise, ultimately, said quickly and summarized, is
that there is not such direct empirical data.

		This is a slide which I think by now is more than familiar to
everybody in the room, and which Mr. Perry indicated had, in fact,
continued to decline in -- at least his comments yesterday, and which
shows a 93-plus percent decline in the rate of silicosis deaths. 
Yesterday there were certain questions about the meaningfulness of the
slide, and I want to talk about it from a slightly different
perspective.

		One of my concerns has to do with the long latency of silicosis, and I
think it's an issue that pertains largely to interpreting the data that
are on the table.  OSHA in its post-rulemaking, or in this case the
background document, that's shorthand for a very long named document,
which is a summary of medical information and a risk assessment, which
was part of the 2010 docket, says that the latency period for silicosis
is 10 to 30 years after first exposure.

		In fact, review of the literature of some of the more important, kind
of classical silicosis case reports indicates that the latency is
actually significantly longer.  The Hnizdo and Sluis-Cremer report of
South African gold miners with 313 cases indicates a mean latency of 35
years and a range of 18 to 50.

		The Vermont granite workers by Costello and Graham reports a series of
38 deaths and a much larger number of workers, and says that 92 percent
had latency greater than 30 years.

		de Klerk and Musk, in Australian gold miners, presents a large series,
with 630-plus cases of silicosis in which the peak incidence of
silicosis was actually 45 years after first exposure.

		A more interesting study from several standpoints is the very large
Chinese cohort study, which continues to be the subject of ongoing
publications as recently as the last few months subsequent to the OSHA
rulemaking.  That's this cohort often referred to as the Chen cohort.

		But there is this one very interesting study from 2006 that speaks to
the latency.  It's published in a journal that I think few of us
actually subscribe to, but which is available, at least through the Yale
Library and probably others, in which a consideration of latency and
progression was made.

		OSHA cites this paper in its documents as having a latency period of
22.9 years, but there's a different side to that if you look at the
context of the exposures.  This is a table that comes from that paper,
and what you can see is that OSHA has taken the weighted average latency
for a period of close to 40 years.

		The left column describes the year of first diagnosis.  And what is
really striking when you look at this is the fact that the latency
described in this paper, over the course of the 30 or so years of the
captured time, tripled, that the typical latency went from about 11 or
12 years to about 33 to 36 years.

		That's striking.  That doesn't have a very obvious biological basis. 
And it's useful to look to see what might have provided such a wide
range of latencies.

		Chen, another paper in that series, this one from '92 in the Journal
of Occupational Medicine, speaks of the facts that operations in this
cohort, this included 20 mines of four types and 9 or 8 pottery
factories, and that operations began before 1949.  

		They included workers who had been exposed for at least one year,
working between '72 and '74, but not those who started in '70 to '74. 
Many of the workers had preceded for quite a long time.

		And in that paper they provide some insights into the levels of dust
to which people were being exposed, and you can see that the ranges in
that study went up to close to 25 or 30 mg/m3.  We're talking two to
three times the OSHA PEL and the TLV for nonspecific particulate.

		And in 2006, in another reiteration and follow-up of that study, they
described something of the time distribution of those exposures, that
during the 1950s until about 1958, when the first exposure controls were
implemented, that the average exposures were in the range of 10 to 25
mg/m3, and at that time they did not use dust protection on individuals.
 There was no personal protection.

		That those levels, starting around 1958 or 1960 fell dramatically, but
they still remained in the area of 4 to 8 mg/m3 -- once again, we're
talking total dust, not crystalline silica, and then gradually
progressed downward.  So we're looking at a rather dramatic high end and
a rather striking decline.

		And I'm so happy, because when I looked at these slides last night on
an iPad, this didn't project, so I'm just a happy camper now.  

		This is a figure that comes from the 2012 Chen study.  This is after
those dust measures have been converted to respirable silica and
crystalline, and the three mines which are the subject, or the three
cohorts which are the subject of the Yang study, which I've just shown
you, are in the purple, the red, and the green.

		And they just serve to emphasize the dramatic step-down and the period
of exposure, that these were not average exposures.  You can average
them, but there was no typical or average.  They were really
time-dependent.

		And what I've done here is to say that if we blocked out the first or
three tranches of the diagnostic group and look at those that were most
recent in the last 15 or 20 years, what you see is that the weighted
latency period for those workers is in the range of 30-plus years, not
the lower period.

		It serves to do two things.  It serves to reemphasize that the average
latency or the mean or the median, however you want to take a central
tendency, is longer than estimated in the OSHA proposed rulemaking
documents, and secondly, I think it provides some interesting insight
into the question of dose rate as opposed to simply cumulative dose.

		I think that it's hard to escape the fact that there was just an
enormous dose rate during those early 8 or 10 years or more, when people
were being exposed to humongously high, extraordinarily high levels. 
And that is probably a component of both the development of disease and
the rate or progression of disease.

		We can also look at the latency issue looking at the SENSOR data. 
This is a point I'm going to make very quickly because it's kind of
reiterative.  David Weissman yesterday already described this in his
testimony from NIOSH, that the SENSOR data really does not pertain to
those cases that are being found nowadays.

		But what I'm looking at here, or showing you here, is a figure that
comes from the most recent 2012 NIOSH publication of the SENSOR data,
and that shows the decades of first exposure of those cases entered into
the SENSOR registry between 1993 and 2006.  The little blue box on the
far right side simply shows you the period of time during which case
ascertainment was being made.

		This shows you what would have been a 30-year latency if all of those
cases had been detected on the first day of that recognition period.  If
I move that out to the midpoint in that recognition period, you see that
there's a very small proportion that have less than at least a 30-year
latency, and 40 years, clearly more.

		And these are, in fact, the data that come from the most recent three
SENSOR publications regarding effective latency and onset of silicosis. 
In the 1997 publication from Michigan, 96.3 percent of the cases
recognized between '87 and '95 had first been exposed before OSHA, in
fact, came about, and clearly before the PEL was enacted.

		In the second series, 90.5 percent, and the most recent -- the one
that I just showed you, distribution, 87 percent of those cases were
exposed before the adoption of the OSHA Act and the promulgation of the
PEL.  

		Simply to say that it is difficult to use those SENSOR data for the
purposes of trying to evaluate the current day risk of exposures,
because the exposures that contributed to SENSOR data predated the OSHA
PEL.

		OSHA has relied upon SENSOR data, and also a couple of other related
studies, Goodwin and Rosenman, to estimate the under-reporting of silica
cases.  And Patrick Hessel will discuss some of the methodological
limitations of those two studies, and discuss some of the limits to
their use for the purpose of trying to extrapolate and estimate
under-reporting.

		Then there's another interesting aspect, which is the long duration of
survival after the diagnosis of silicosis.  This is -- well, there were
a series, I think the white progresses down with progressive slides, a
large study of Italian silicotics actually compensated for silicosis,
1258 deaths, and the survival time between first diagnosis and death was
a median of 23 years in that series -- large series.

		In the Yang study, which I mentioned a bit earlier, the survival time,
the mean was 21.5 years from diagnosis of silicosis 1 on x-ray, to the
time of death, and 25 percent were longer than 33 years.

		In the pooled IARC study of silicosis, with 170 silicosis deaths,
while there is not a specific, clear statement of the duration of
survival, what you see is that the median duration of exposure was
28 years, and only 9 percent died within a year thereafter, so it's
clear that there was a combination of exposure and duration that would
have been quite a long period of time.

		The implications of the latency and the survival durations are fairly
straightforward, going back to that familiar CDC chart.  If you consider
that a mean latency of 30 years and a mean survival of 20 years -- and
I understand that that's a simplification, but for the moment, if you
assume it, then the fact is that virtually all the cases reported as a
mortality were first exposed in a period that preceded the adoption of
the PEL and the adoption and enactment of the OSHA Act.

		And it suggests that the data have only limited probative value with
respect to whether this speaks to the permissible exposure limit and its
adequacy in protecting workers.

		There are obviously cases of silicosis that are associated with
shorter latency, and I don't mean in any way to ignore those.  Acute
silicosis, a horrific disease which occurs with very, very high
exposures, and which can have an onset in a matter of even weeks, but
certainly within a matter of only a few years, described by OSHA as
being intense exposures, high crystalline silica content, very high
concentrations.  

		And accelerated silicosis is generally regarded as being an onset
within 5 or 10 years, again to very high exposures, intense, often
unprotected, substantial overexposures, says OSHA, and extremely high
exposures in the words of NIOSH researchers, which simply serves to say
that to the extent that there is short latency onset of silicosis --
and there is, it does not provide evidence of the PEL's adequacy or
inadequacy, because those short onset cases are almost all exposed to
these extremely high levels, which do not provide insight into a
proposed or enacted 100 µg/m3 PEL.

		So in summarizing this piece, I think that neither the SENSOR data nor
the remaining silicosis-related deaths in the CDC data demonstrate that
the current PEL is not protective, simply because there is high
exposures that predated OSHA, because the shorter latency cases due to
substantial overexposure and unprotected don't count as evidence against
the current PEL.

		And so I think the SENSOR data and the CDC death data actually are not
a factor in the assessment of the adequacy of the current PEL.  

		There is also significant evidence from OSHA and others that there is
and has been extremely widespread overexposure to silica despite a
significant and steeply declining trend in exposure levels.

		This is a composition of data, mostly OSHA data.  The first six lines
are from a study by Yassin and colleagues, OSHA scientists looking at
the Integrated Management Information System, a compilation of OSHA
compliance data, and the latter two are German data.  The first one,
actually, is cited in the OSHA preliminary analysis and the second one
is not cited therein.

		And what they show are the levels of exposure in the duration of '79
to 1987, and then the mean between '88 and '03.  And the third column
shows you the extent of the decline in the silica mean exposure levels
over those time periods.

		And what you can see based upon those data, that the decline that did
take place was between -- in those industries, 50 to 85 or 90 percent,
a rather dramatic decline.  Nevertheless, it did not eradicate a
persistent and important overexposure, which I think is something that
needs to be addressed and hopefully will be.

		In the CDC publication, which included the original of that
now-familiar 93 percent decline, CDC said that despite the reductions in
mortality, silica overexposure remains widespread.  The NIOSH
scientists, commenting upon the persistence, said that despite the
existence of enforceable limits on worker exposure to respirable
crystalline silica, substantial exposure continues to occur in the U.S. 
That's in 2008.

		And the OSHA background document cites a range of 0.12 to 0.48 --
that is from above the current PEL to above four times the current PEL,
and says that that falls within the range of exposures of particular
interest.  Exposures in this range or not much lower are common today.

		Simply to say that there is a persistent and important ongoing
overexposure, which is clearly a health concern, and which needs to be
addressed.

		Here are some data that reflect specifically on the extent of those
ongoing and current overexposures.  This is from NIOSH 2007, which says
that from '93 to 2003, about 23 percent of the silica levels in coal
mining, 11 percent in metal/nonmetal mining, and approximately 40
percent in the following industries were over the PEL.

		And they go on to speak about iron and steal foundries, construction
machinery except electrical fabricated structures, with at least
one-third of their exposures exceeding the PEL.  These are recent
data -- or not most current data, but these are fairly recent.  These
are not historic.

		These are data that come straight out of the preliminary economic
analysis.  These are the data that were largely but not entirely put
together by Eastern Research Group, and they give a cross-section of
some of the worst offenders in the industries and sectors.

		Looking at the proportion of data that were over the current PEL,
that's the far right hand, or over 250 µg/m3 -- that's the
intermediate or the middle column -- and what you see going through
this is that a very significant proportion of the data cited by OSHA as
current or recent is largely above the OSHA PEL.  And it's a dramatic
and important degree of exceedances, which needs to be addressed and
should be the focus of attention, I think. 

		In a series of ongoing studies and others which contribute additional
insight, Mary Ellen Flanagan and her colleagues at the University of
Washington looked at a large number of construction workers, 13
databases including the OSHA IMIS database, 1375 or so personal samples,
found that 13 percent of all of those samples exceeded 1 mg crystalline
silica, and went on to caution that the range of exposure for any tool
or task is extremely broad, spanning three or four orders of magnitude
for most tasks or tools.

		This wide range means that using any measure of central tendency to
represent an activity can seriously underestimate exposures in some
cases.  My point here is to emphasize the extent of the overexposures
that continues to persist in American industries.

		This is a graph from a paper by Stephen Rappaport.  He looked at four
different categories of construction workers.  The red line is the OSHA
PEL.  The black line is the NIOSH REL.  And we're looking at these four
different groups of workers who represent a very large portion of the
construction work industry.  And what we're looking at is a tremendous
number of exposures which exceed the PEL, and even more exceeding the
REL.  

		The bottom line seems really simple to me.  It is that there is a
gigantic amount or risk of a very large pool of people who are being
significantly overexposed, and that that database is so great and so
pervasive that it makes evaluation of the PEL as a -- and its adequacy
to be protective almost moot, because it's very hard to attribute any of
the cases being recognized to people who are being exposed at or even
near the PEL.

		Let me remind you of this slide which I showed earlier.  Right now I'm
showing it to you for one very simple reason.  I want you to look at the
left-hand column and just see how bad exposures were in the period of
'79 to '87.  I don't mean to beat the drum on this, but I am going to.

		The point is that if you allow for a 20- or 30-year latency or even
less, as recently as 20 or 25 years ago there were workers who
apparently were being regularly exposed to levels that are just
egregiously high.  And their impact on the pool of silicosis or
silicotics or potentially such people is obvious.  I mean it's really a
rather remarkable thing.

		The OSHA Preliminary Economic Analysis data reflect recent but not
current exposure levels, and it's unclear to what extent that a
significant decline in exposures has continued.  ERG primarily looked at
data from the late '80s to 2001.  OSHA updated a little bit, they say
through 2007, and it's true.  

		But the fact is that there's at least a strong trend of decline, and
it's interesting but not documented in the OSHA materials as to whether
that decline has continued.  And I understand but don't know on first
hand that the data that would provide such insight has not been made
generally available.

		Then let's jump from that to issues of the risk assessment.  OSHA, I
think, correctly says essentially that the CDC and SENSOR data are not
sufficient to be able to draw conclusions with regard to the
occupational exposure limit for crystalline silica, and that it is the
dose-response data that are appropriate for that purpose, the
hypothetical risk assessment applications rather than the empirical data
itself.

		OSHA goes on to say that it believes that uncertainty in the exposure
assessment that underlie each of, for example, the 10 studies in the
IARC pooled meta risk assessment represent one of the most important
sources of uncertainty in the risk assessment, and the authors of that
paper concur.

		Nine of the 10 pooled studies used exposure methods that differ from
those in the proposed Notice of Proposed Rulemaking, and go on to say
that:

		"Measurements in all of these various different units and using all of
these different methods require conversion to a common unit for purposes
of conducting a risk assessment.  And one weakness to our data is our
reliance on conversion factors for which, generally, we have no way to
assess the validity."  It's a concern to the underlying nature of that
risk assessment.

		"In addition, the purpose of the analysis was to determine exposure
response trends, not point estimates of risk levels."  And the
cautionary note was that categorical analyses based on broad exposure
groups could be performed using those data.

		OSHA considered the impacts of some of those exposure
misclassifications, particularly the data as described in the toxic
chemical report, use of job-specific average exposure levels, use of
average conversion factors for dust to respirable crystalline silica,
and use of biased conversion factors in the one South African study.

		And Dr. Chris Long -- who's on my right, one over, will detail
numerous other sources of exposure misclassifications and errors and
likely impact on the respirable crystalline silica risk assessment.

		There are also, in addition, concerns over and beyond those of the
exposure misclassifications that have to do with errors in the exposure
measurements themselves.  There's known variability and quantifiable
errors associated with sampling pumps, samplers, laboratory analyses. 
And Dr. Tom Hall will discuss -- and Dr. Robert Lieckfield will
discuss this afternoon the variability and errors that characterize
OSHA's proposed sampling methods.

		They will also talk about the magnitude of those errors and raise
concerns about the feasibility of the OSHA-proposed methods for the
purpose of monitoring at the proposed PEL and action level, and the
capacity to meet OSHA's own proficiency standards for accuracy and
precision.

		Now, sampling and analytical errors have not been incorporated into
the risk assessments to the extent that the actual measurements cited in
the text have all been treated as if they were error-free.  Chris Long
will discuss the effects of sampling, analytical, and other measurement
errors on the risk assessment, including bias of effect estimates,
distortion of exposure response curves, and the obscuring of estimates
of threshold concentrations.

		The issue of thresholds is something which is really understated and
needs to be highlighted and dealt with, I think, with more detail in the
final documentation that comes out of OSHA.

		In addition, I point out that the adoption of the ISO/CEN convention,
which provides the advantage of harmonization, which is, in and of
itself, a desirable thing, results in a greater mass of particles
collected for any given exposure level.

		And OSHA says for most workplace conditions, the proposed change to
criteria would increase the mass of respirable dust collected over that
measure over the current criteria.

		And Soderholm has addressed the extent to which that will have an
upward bias on the loading that results and the implicit exposure
levels.  And the problem will be that the exposure levels that are taken
using the ISO/CEN convention will be inconsistent with the exposure data
that underlie the risk assessments.

		The risk assessments have not incorporated the fact that the ISO/CEN
will change the amount of loading and the implicit levels of exposure
that will result from the use of the new pumps.

		Now, the definition of respirable crystalline silica by OSHA, airborne
particles that contain quartz cristobalite and/or tridymite, whose
measurement is determined by a sampling device designed to meet the
characteristics of the respirable size selective samples, is ambiguous,
unfortunately, although we all understand what is intended.

		Because the performance of individual samples, as Tom Hall will tell
you later, vary, the proposed performance-based definition would yield
variable results depending upon the sampler used for any given
environment.  And that will be something which will be discussed during
the afternoon presentation.

		Recommendations to simplify and avoid the ambiguity of that definition
include, one, defining respirable crystalline silica as the quartz that
is contained in the air and not as the particle containing quartz that
is in the air, and to define respirable crystalline silica by size
rather than performance standards of a sampler, given the fact that the
samplers are all somewhat variable and prone to under- and over-sampling
in various and idiosyncratic ways -- not idiosyncratic, predictable but
distinct to each sampler.

		Exposure response thresholds, there is substantial support in the
literature for the view of exposure response thresholds between inhaled
silica and the development of silicosis, that is a threshold function. 
And Chris Long will speak to that particular concern and how exposure
measurement errors can contribute to obscuring the exposure response
thresholds.

		He's going to speak, Chris will, with regard to silicosis.  IARC, in
its two monographs on silica, have made the point that there is no
convincing evidence for a direct physical-chemical mechanism for
crystalline silica-induced genotoxicity.  And the question becomes, is
there a direct mechanism of silica-induced lung cancer?

		IARC summarized seven meta-analyses in their most recent monograph and
concluded that there were increased risks of lung cancers among workers
with silicosis, and no increased risk of lung cancer in those without
silicosis.  

		I think there's no question in people's minds that silicosis itself
has a threshold.  So the question of whether silicosis is a necessary
precursor becomes a very important aspect of doing the risk assessment
for lung cancer itself.

		There is a very interesting recent paper, I'm sure you are all aware,
by Liu et al., published last summer or early fall and included -- not
included directly in the Notice of Proposed Rulemaking, but included in
any number of the comments that have been submitted, which claims to
break new ground by documenting a substantially increased risk of lung
cancer in those Chinese cohorts of whom we spoke earlier -- I spoke
earlier, who did not have silicosis.

		But the absence of silicosis in that study was defined as negative
chest x-rays.  And it is important to consider -- and surprisingly,
those authors did not -- that there is a very direct dose-relatedness
between level of exposure and probability of false negative chest x-rays
for silicosis.

		Now, that particular set of data comes about from a series of studies
that Eva Hnizdo and colleagues did in 1993 in South Africa, in which a
series of workers, underground miners who were in a healthcare program,
unselected otherwise, and those who went to autopsy were evaluated.  

		Of the 500-plus individuals with silicosis in their autopsies, then
pathology studies, their pathology was compared to their most recent
chest x-rays.  And the data showed that the prevalence of or the
prediction of the chest x-rays for negative values declined sharply. 
The probability of a false negative increased dramatically in a
dose-related way.  

		This is one of the tables from that study.  What we're looking at
there is, on the vertical, the probability of a false negative chest
x-ray, that is, a chest x-ray without silicosis, in a worker
subsequently found at autopsy to have silicosis.  

		And what we're looking at along the x-axis is the duration of
exposure.  And those five lines going across reflect different levels of
dust exposure ranging from 100 to 500 micrograms to milligram per cubic
meter.

		And, in fact, when you look at that report, what you see in the
highest exposure group of the Hnizdo cohort, that false negatives --
that negative x-rays with false negatives in the highest exposure group
up to 92 percent of the time.

		That very strong dose-relatedness of false negative x-rays is not
considered in the context of the Liu study, and I suspect -- but it's
beyond my mathematical skills, so let me make that clear, that if one
were to combine the estimates of false negatives in that study with the
actual data, you would find that there was little or no increase in the
risk of lung cancer in those with negative chest x-rays, because in
those groups with positive associations, there would be very high
numbers of people with normal chest x-rays and positive silicosis.

		So I'm concerned that this study is going to be over-read by others,
and so I raise that as a note of caution.  

		These are the data, just to indicate again, the levels of exposure --
this is to respirable crystalline silica, in that underground -- in
that study of the Chinese workers, the pottery workers and the three or
four groups of underground miners.

		Interesting to note that the iron and copper workers in the lowest
tier, who would be the group most relevant to evaluating the 100 µg/m3
PEL, were not independently studied.

		In summary, I think that there is a lack of direct evidence that the
100 µg/m3 standard is not protective.  I think that there are no
empirical data on workers at that level.  

		There has been and continues to be widespread exposure and
noncompliance, and as a consequence there are lots of workers who
unfortunately are being overexposed, and they are being overexposed in
many of the industries most likely to be associated with silicosis and
silica-related health effects.

		The exposure assessments have been uncertain for a variety of reasons,
including the technical limitations of the measurement sampling and
analysis methods.  And also because of the ways in which the existing
data have been used, there has been a tendency to discount the exposure
response threshold, which are almost certainly true for silicosis and
most probably true for the lung cancer.  Silicosis is a necessary
precursor, and the definition of respirable crystalline silica is
probably too ambiguous, and it will be problematic.  Thank you.

		JUDGE SOLOMON:  Okay.  We've used up approximately half the time that
was allotted, so you also have some time this afternoon.

		MR. CHAJET:  We do.

		JUDGE SOLOMON:  So how can we use, best use, most efficiently use the
rest of your time this morning?

		MR. CHAJET:  I think we should continue to go through the panel.  We
will have, I believe, enough time left to finish the panel.  And if we
combine the time from this afternoon and this morning, I'm sure we'll be
able to complete it.

		JUDGE SOLOMON:  Okay.  Go ahead.

		DR. LONG:  My name is Chris Long, and as Henry mentioned, I'm a
principal scientist at Gradient, an environmental and health science
consulting company based in Cambridge, Massachusetts. 

		In the next 10 or 15 minutes or so, I'll be focusing on exposure
measurement errors as potential sources of biased risk estimates, and of
threshold obscuring or threshold shifting effects in the silica worker
epidemiologic cohort studies that are the basis for OSHA's risk
assessment.

		Exposure measurement error, which can be defined quite simply as the
difference between an estimate of worker exposure that is used in an
epidemiology study and an actual personal exposure level, is recognized
to be a common source of uncertainty in most occupational and
environmental epidemiologic studies.

		This is particularly the case for the silica worker cohort studies,
given the general scarcity of exposure measurements across all affected
individuals and exposure time periods of interest.  Both OSHA and its
peer reviewers have acknowledged exposure measurement error to be an
issue of significant importance to OSHA's risk assessment.

		Some of the issues that I'll be discussing will not be new to the OSHA
peer reviewers, since they've also raised them.  

		I've organized my comments on exposure measurement error around three
main points.  First, it is now well recognized that exposure measurement
errors can have significant impacts on occupational and environment
epidemiologic studies, with impacts including effect estimates that are
biased high, distorted exposure response curves, and either complete
masking of threshold concentrations or underestimation of threshold
concentrations.

		Second, the silica worker cohort studies are subject to the potential
impacts of a variety of different sources of exposure measurement error.

		Third, OSHA's assessment of exposure measurement error is very
limited, in terms of the sources of exposure measurement error that it
considered, and in its adequacy for addressing questions related to
potential impacts of these errors and risk estimates and thresholds of
response.

		So I'd now like to make a few general comments about exposure
measurement error and its well-recognized ability to bias regression
coefficients, modify exposure response relationships, and either hide or
cause an underestimation of threshold estimates.

		These general comments are based in part on a review paper that I
co-authored, along with several other Gradient scientists for
publication in the peer-reviewed journal, Critical Reviews in
Toxicology.

		This paper, which is titled, "Measurement Error in Environment
Epidemiology and the Shape of Exposure Response Curves," was published
in 2011.  As indicated in the title of the paper, it focused on the
existence and impacts of non-differential exposure measurement error on
the shape of exposure response curves and the estimation of threshold
parameters.

		As discussed in great detail in the paper, it is now well known among
statisticians and epidemiologists that random errors in the values of
independent variables, such as estimates of exposure, can bias
regression results.  This is true for random errors, including both
Berkson type and classical type random errors, as well as systematic
errors.

		Moreover, there are several examples in the literature of how this
bias can sometimes be sufficient to distort the shape of exposure
response curves, and to either completely obscure true thresholds or
result in threshold underestimation.

		Based on our comprehensive review of the epidemiologic and statistical
literature, we concluded in the paper that error in independent
variables, such as exposure variables, tends to smooth out and flatten
and essentially linearize curves from a variety of different types of
regression models.

		Written comments submitted to OSHA provide several examples from the
occupational and environmental epidemiology literature, but I'd like to
highlight one example that I find to be particularly compelling.  

		It's from a 2002 paper from the journal, Risk Analysis, titled
"Exposure Misclassification and Threshold Concentrations in Time Series
Analyses of Air Pollution Health Effects."  That was published by
Michael Brauer and several other researchers at the University of
British Columbia.

		The Brauer et al. paper focused on Berkson type random error arising
from the use of community average PM2.5 data from outdoor central site
monitors to represent personal exposures to ambient PM2.5.

		I think important lessons can be drawn from this scenario, where
community average PM2.5 data are used to represent individual personal
exposures for the common scenario in the OSHA risk assessment where
group level average respirable crystalline silica exposure estimates are
assigned to individuals in silica worker studies.

		Brauer et al. conducted simulations where individual level mortality
risks were estimated for exposure response models incorporating the
source of exposure measurement error in different threshold assumptions.

		As persuasively demonstrated by figures in the paper, there is no
evidence of a threshold at the lowest value of the true threshold
parameter.  While the exposure parameter, while the exposure response
curves for the other, higher assumed threshold levels retained some
evidence of non-linearity, the threshold estimates were biased low.

		As we noted in our paper, these findings strongly suggest that the
inability to detect a threshold in many epidemiology studies does not
mean that no threshold exists.

		Next, I'm going to move on to exposure measurement error specific to
the epidemiologic studies of silica-exposed workers.  You know, we
discuss in great detail in our written comments to OSHA how these
studies are likely subject to a number of additional sources of
potential exposure measurement error beyond the three error sources
considered in the assessment of exposure measurement error in OSHA's
report.

		You know, OSHA's analysis considered three sources of exposure
measurement error, namely two sources of Berkson type random error,
specifically the assignment of job-specific average exposure levels from
a job exposure matrix, and secondly the use of average conversion
factors to convert historical dust measurements to gravimetric
respirable silica equivalents.

		In addition, they considered a single source of systematic error,
namely the use of biased conversion factors for converting historical
dust measurements to gravimetric respirable silica equivalents.

		However, in our written comments, we've identified at least 10
additional sources of potential exposure measurement error in a
comprehensive table.  I'm not going to describe each of these additional
sources of exposure measurement error, but a few prominent examples
include the following:

		First, error associated with the estimation of historical levels,
which necessarily results from a general lack of exposure measurement
data for many worker cohorts for time periods prior to the 1950s. 

		Second, error from between measurement variability, such as due to
factors such as instrument error, sampling and pump error, which will be
talked about in more detail this afternoon.

		Third, error from between sample duration variability; this can arise
from the use of short-term samples versus full-shift samples.  

		Next, error from between sample location variability; this can arise
due to the use of area samples versus personal exposure samples.

		Also error from between sample type variability, due to the use of
measurements targeted to assess compliance versus measurements from
random exposure surveys.

		Following on this comment, I'd now like to briefly discuss some of the
key shortcomings of OSHA's assessment of exposure measurement error that
limit its utility for making any conclusions regarding the role of
exposure measurement error as a source of bias to silica effect
estimates and threshold estimates.

		These key limitations include the following:  First, as I just
discussed, the estimate only considered a small fraction of probable
error scenarios for these kinds of studies.  Second, there are no formal
analyses conducted to determine the error structures of the three
sources of exposure measurement error included in the sensitivity
analyses;	for example, without any formal analysis, the OSHA assessment
simply assumed a purely Berkson type error structure from the assignment
of job-specific average exposure levels for individual exposures. 
However, there are a number of published studies that have shown that
this type of exposure measurement error can often have a more complex
error structure with properties of both Berkson type and classical type
random error.

		JUDGE SOLOMON:  While he's taking a break, I neglected to say at the
beginning -- I did say it yesterday, that anybody who has a cell phone
or any other device should make sure that it's off.  And if there's any
conversation, if you need to talk to each other, we would appreciate it
if you go outside and do it.  Go ahead.

		DR. LONG:  Thank you.  Next, another limitation the OSHA assessment
was its assumption of log linear -- it assumed essentially specific
types of models, including log linear models with log-transformed
exposure variables, and it focused on cumulative measures of silica
exposure that obscure both within-person and between-person variability
in exposure rates.

		Just by limiting itself to these types of models, it's just, it's not
known whether other possible models and exposure metrics, such as
non-transfer in exposure variables, are more or less sensitive to
exposure measurement error.

		Another key limitation involved its narrow focus on impacts to
regression coefficients rather than impacts to the shape of exposure
response curves and threshold estimates.  There's very little discussion
in the OSHA report regarding the potential impacts of exposure
measurement error on identification of thresholds.  And this is true
despite peer reviewer comments noting how little attention is given to
the threshold issue.

		Moreover, ToxiChemica, which is the group that performed the
sensitivity analyses that are the centerpiece of OSHA's assessment of
exposure measurement error, noted that exposure response threshold
estimates are imprecise and appear to be highly sensitive to measurement
errors.

		Next, the sensitivity analyses that were conducted by ToxiChemica for
the OSHA assessment were restricted to studies of lung cancer and
silicosis mortality and did not address studies for other health
endpoints including silicosis morbidity and renal disease mortality. 
It's thus unclear whether data for these other silica-related health
endpoints are more or less sensitive to exposure measurement error.  

		For lung cancer and silicosis mortality, OSHA includes a rather uneven
discussion of the mixed and inconsistent findings of the ToxiChemica
sensitivity analyses that fails to adequately discuss their
implications.

		For example, OSHA concludes that exposure measurement error is
unlikely to be a major source of uncertainty to the estimates of lung
cancer risk, despite finding that a few cohorts were highly sensitive to
the two simulated sources of Berkson type exposure measurement error.

		In particular, one of the studies of South African gold workers, there
was a 70 percent reduction in the effect estimate once the two sources
of Berkson type random error were accounted for.

		In addition, OSHA noticed that the silicosis dataset was found --
silicosis mortality dataset was found to be more sensitive to the
impacts of the three sources of exposure measurement error than the lung
cancer dataset, but offers no discussion of the reasons for this
difference.  Strangely, for silicosis mortality, there really are no
clear conclusions on the significance of the exposure measurement
errors.

		To conclude, I'd like to emphasize that our written comments delve
into these topics in greater detail.  Rather than repeating each of the
conclusions from our written comments, I'd like to highlight two key
conclusions.

		First, overall, is our conclusion that the silica worker cohort
studies are subject to a variety of potential sources of exposure
measurement error beyond just the three types considered in OSHA's
assessment.  Due to this limitation as well as other limitations that
I've discussed, OSHA's assessment is highly limited and insufficient for
ruling out potential biases from exposure measurement errors that can
affect the interpretation and use of exposure response associations from
the epidemiologic studies of occupational exposures to respirable
crystalline silica.

		Accordingly, OSHA fails to provide adequate scientific justification
for the conclusions of its risk assessment.  There's a need for OSHA to
conduct additional analyses to further assess the potential impacts of
exposure measurement errors on its risk assessment findings.

		In particular, you know, this is true with respect to the
identification of possible thresholds of response.  It may be prudent
for OSHA to reanalyze the epidemiologic data using available techniques
for correcting for potential biases from exposure measurement error,
rather than treating the literature exposure estimates as if they're
accurately known and error-free.

		Thank you for your attention.  I'd like to turn it over to my
colleague, Dr. Peter Valberg.

		MR. CHAJET:  Then I guess we should continue.

		JUDGE SOLOMON:  We are going to take a break, but I'll wait until
you're done with the -- with your presentations, unless anybody has
some urgency.  Okay, go ahead.

		DR. VALBERG:  All right, thank you.  My name is Peter Valberg, and
I've spent a fair amount of time in the area of human health risk
assessment, and I'd like to offer a few comments to OSHA with regard to
the Preliminary Quantitative Risk Assessment.

		A key question in the nature of exposure is response relationships, is
the nature and shape and functional form.  For inhaled crystalline
silica, the existence of a threshold -- and by threshold, we mean a no
adverse effect level -- is suggested by several lines of evidence,
including animal toxicology, mechanistic considerations, and some of the
epidemiology studies.

		And as Chris has already mentioned, the peer review panel of 2010
offered some interesting advice pointing to the need for careful
consideration of thresholds, and yet it seemed like the Preliminary
Quantitative Risk Assessment didn't really follow up on this particular
advice.

		For poorly soluble particles like crystalline silica, biological
mechanism and laboratory animal evidence support the idea that
persistent lung inflammation is a necessary step in the development of
both malignant -- both nonmalignant and malignant lung disease.

		A persistent inflammatory state can develop when inhalation exposure
exceeds the ability of the lung to clear deposited particles and to
neutralize free radicals and cytokines that are released by cells in
response to particles retained on lung surfaces.

		We all inhale particles on a daily basis, and our lung defense
mechanisms are capable to deal with these.  And if those lung defense
mechanisms are not overwhelmed, the lung will be able to deal with
particle burdens without becoming diseased.

		In fact, in a recent 2014 article, Dr. Kyle Steenland made the
comment that while there's also some low-level silica exposure on
beaches and in ambient air in general, there's no evidence such
low-level exposure causes effects.  

		Thus, it's important to consider or to test those response models that
include the possibility of exposure concentrations of respirable
crystalline silica below which there are no anticipated health effects,
the so-called no adverse effect level.

		And as Chris has already emphasized, the model choice affects this a
great deal, because even if the biological response does have a well
defined no-effect threshold, failing to explicitly recognize and deal
with errors and uncertainties in the assignment of exposure can easily
cause the linear component of the model to dominate and obscure the
threshold.

		Now, in OSHA's Quantitative Risk Assessment document, it appears that
the linear and loglinear exposure response models were the focus.  And,
in fact, on Page 269 of the QRA document, OSHA identifies exposure
response models that it used, and the studies that it selected in the
Quantitative Risk Assessment.

		None of them seemed to test, as far as I could tell, a threshold
model.  On Page 290, OSHA also combines all of these into a table, which
indicates that the emphasis was on linear fitting.

		In fact, OSHA seems to specifically dismiss the idea of a non-linear
model by saying, on Page 272 -- and here I quote from the document
itself, "OSHA believes that the lack of a consistently monotonic
exposure response curve from the spline model lessens its credibility,
since there's no reason to believe that the biological response to
increasing silica exposure should not be monotonic throughout the range
of exposure."	

		So the question is, is that really the case?  And I think if we look
at, a little bit, at some of the evidence that can be applied here,
we'll see that there are definitely very strong suggestions that
thresholds do occur.

		In terms of the animal evidence, I think if we look at the rat lungs
which are often used in these kinds of experiments, we see that their
response to so-called poorly soluble particles, of which silica is a
type, they can reach a lung condition called lung overload that is
characterized by persistent inflammation.

		This lung overload state, as far as I can tell, is not discussed in
the OSHA document, and it's characterized by neutrophil influx,
macrophage influx, oxidant and cytokine release, and increased pulmonary
epithelial cell hyperplasia.

		And these kinds of responses represent a generic, so-called epigenetic
mechanisms, whereby the rat lung responds in an exaggerated fashion to
particle inhalation.  As exemplified by the rat model, the triggering of
persistent inflammation is a key threshold element in respiratory
disease and tumorigenicity.

		For exposure levels below the threshold needed to trigger this
persistent inflammatory state, retained particles do not lead to
irreversible adverse biological effects.  They may trigger lung defense
mechanism enhancement, but not irreversible effects.

		Specifically, in an article that is cited by OSHA, Porter, they noted
that the response of the rat lung to inhaled crystalline silica
particles is biphasic, with a below-threshold phase characterized by
increased but controlled pulmonary inflammation.

		They concluded -- and here I quote, "Pulmonary defense mechanisms
were initially able to compensate and control silica-induced pulmonary
inflammation and damage, but after a certain threshold lung burden was
exceeded, these control mechanisms no longer were adequate to prevent
the progression of silica-induced pulmonary disease."			And so it's
these biological mechanisms that need a little more consideration in the
OSHA document.  In fact, Klein and Christopher of the California
Department of Toxic Substances Control reviewed data on crystalline
silica fibrogenicity and carcinogenicity and determined -- and here I
quote from those authors, "The safe level of silica should be based on a
threshold model in which fibrotic lung disease is a necessary
precondition for the appearance of lung cancer."

		That is, these authors recommended that the linearized model not be
used to describe the lung's response to crystalline silica inhalation
because the fibrogenesis appears to be governed by a threshold response.

		Now, there are other papers and literature such as that of McHoney et
al., who suggested that pneumoconiosis following exposure to
silica-containing coal dust is more likely to be caused by iron content
triggering a persistent inflammatory state rather than the silica
content itself.  And I think that the peer review panel also suggested
that the mode of action needs to be more clearly fleshed out in the OSHA
document.  

		The OSHA peer review panel raised that in the following way, and here
I actually -- I'm cribbing from the peer review panel, and the quote I
chose here was, "What non-linearities in the biological response might
exist that would be sensitive to dose rate, threshold phenomena such as
exceedance of defense mechanisms or damage requirement needed for
excitation of immune inflammatory processes?"  So it's these kinds of
mode of action considerations that need to be included.

		In fact, as far as I could tell, in the 480-plus-page document, it
appeared that the only place where threshold in the worker epidemiology
studies was discussed was on Page 275, and no specific threshold
study -- threshold testing of any epidemiology study was presented, and
I couldn't find that in the remainder of the document. 

		And in terms of the animal threshold values, the threshold value
derived by the Kempel et al. investigators, through a rat-human
comparison, OSHA looked at that and determined that they wanted to
dismiss the implication, starting on Page 329 -- and here I'm quoting
from the OSHA document, "However, OSHA does not believe that the
analysis of Kempel et al. is definitive with respect to a threshold for
silica-related disease."	

		Now, I think that there is this indication that threshold is being
disregarded, and here again I will crib from several comments made by
the peer review committee, because I think they phrased the criticisms
very clearly.

		One comment was, "The appropriate model should be, would need to be
selected, including the likelihood of a threshold, at least for
non-cancer effects."  Another comment said, "The lung burden model has a
potential utility as it defines a threshold for lung cancer; however, it
is not explained well and this concept is not used."

		Another comment is, "The issue of threshold is not given proper
attention."  Yet another comment is, "More consideration needs to be
given in the document to model selection and its effect upon risk
estimates."

		Another comment is, "The estimate of risk has not adequately dealt
with uncertainty regarding the inexistence of thresholds for cancer and
non-cancer endpoints."  And, finally, the last quote, "The issue of
possible thresholds should be systematically evaluated for each study."

		Now, in terms of a few other papers, in terms of the mode of action, I
mean, OSHA does not cite a key paper by Borm and colleagues.  I think
the date of that was 2011.  The title of the paper was "The Carcinogenic
Action of Crystalline Silica: A Review of the Evidence Supporting
Secondary Inflammation-Driven Genotoxicity as a Principal Mechanism."

		And in reviewing the data, these authors concluded that inflammation
is a driving force for genotoxicity, and that the genotoxicity of silica
particles would play only a role when you had very high overloading
concentrations, exposures, and deposited doses.

		They also noted, this is very much in agreement with what Dr. Borak
said in terms of IARC's discussion of crystalline silica where they
considered the established mechanistic events as being impaired particle
clearance and -- leading to macrophage activation and persistent
inflammation.

		And in his same review paper of 2014, Dr. Steenland also appears to
endorse this idea, saying, and here I quote, "Both silicosis and lung
cancer are believed to result from the strong inflammatory response that
silica evokes in the lung."  And so there the idea is that whether or
not the inflammatory response can become persistent is going to be
governed by threshold considerations.

		There are -- the QRA document does not sufficiently discuss and
emphasize that some worker epidemiology studies suggest threshold dose
response.  The implication of such evidence needs to be more carefully
explored and considered by OSHA's analysis.

		There was a study by Bohr and Tran, which reviewed quartz exposure in
coal mine dust, and they concluded -- and here I will quote from their
abstract, "A no observed adverse effect level for quartz between 30 and
130 µg/m3 over a 40-year exposure is derived.  It is concluded that
more refined physiologically based pharmacokinetic modeling is needed
for a better estimate," that is to say, to narrow down that range.

		And the OSHA document does discuss the work of Graham and colleagues,
where they studied 2500 Vermont granite workers and how the exposure
related to silicosis development.  But those authors noted that there
were no silicosis deaths in workers hired after 1940 who worked only in
that Vermont granite industry, illustrating the effect of the lowering
of quartz exposures.

		Those authors concluded that a no effect threshold exposures were
likely and that they were in the vicinity of 100 µg/m3 for respirable
silica, and that rigorously observed workplace conditions would prevent
the development of radiographic abnormalities.

		Now, OSHA, of course, was aware of that, but they chose to rely
primarily on the reanalysis of Attfield and Costello, who looked at that
same data and did observe a statistically significant trend when
cumulative exposure was log transformed, but not for untransformed
exposure.  And those same authors also eliminated some of the highest
exposure groups to enhance the dose-response behavior.

		There are a couple of other studies, Pukkala, which I think is
mentioned in the OSHA document, sought to understand alternative
exposure metrics for crystalline silica by examining 43,000 lung cancer
cases in the Finnish Cancer Registry and relating them to the Finnish
Job-Exposure Matrix database.

		The authors noted that the stability in most occupations in Finland is
so high that cross-sectional information on workers' occupation
corresponds rather well to his or her lifelong occupational history.  

		And their analysis of possible relationships between crystalline
silica concentrations and lung cancer risk found evidence for a
threshold, in that lung cancer excesses were limited to exposure
concentrations of at least 200 µg/c3.

		They concluded, and here I quote, "The threshold approach may provide
some additional information on the shape of the crystalline silica
exposure response curve."  Now, OSHA does cite to this document, but
they don't really make any reference to the authors' conclusions
regarding the possibility of a threshold.

		Another more recent paper is that of Garber and colleagues, where they
recently published a reanalysis of respiratory disease mortality in
9,000 miners who were enrolled around 1970 and followed up to about
2007.  

		They tested for an exposure response to crystalline silica
concentrations but didn't find any statistically significant
associations.  They did find some statistically significant associations
with total dust, generally, which in that case was coal dust.

		And, finally, I think many people are probably aware of the recent
analysis of Morfeld and colleagues which, you know, was probably was not
in the current document because it was published in 2013, and they
considered approximately 17,000 workers from porcelain manufacturing
plants in Germany.

		These workers were screened for silicosis between 1985 and 1987 and
then were observed for mortality and silicosis morbidity through 2005. 
Respirable quartz exposure was estimated by combining detailed
individual employment histories with a job exposure matrix on more than
8,000 in historical industrial hygiene measurements.

		The authors estimated a threshold concentration of 250 µg/m3, on the
basis of their analysis, which was also the best fitting model.  That
is, these findings by Morfeld and colleagues in the German porcelain
industry show a silicosis no-effect threshold, where below silica
exposure concentrations of 250 µg/m3, the lung responses did not
progress to silicosis.

		So, in summary, the bottom line is that the OSHA exposure response
modeling focuses on linear no threshold models, and therefore they
discovered such a linear relationship.  And that, however, would be true
even if the response were a threshold one.

		The exposure misclassification, as Dr. Long has mentioned, will lead
to a non-linear -- non-zero linear coefficient even when any
epidemiology data are run through this data.

		Now, as has been recognized in publications by Dr. Tony Cox, the peer
review panel, and others, exposure misclassification error can obscure
and misidentify thresholds in exposure response.  Care must be taken to
analyze worker data for respirable crystalline silica exposure in a way
that allows for some possible threshold behavior.

		Finally, neglecting to include exposure response models in the
uncertainty analysis allows you to make a statistical significance
determination that may in fact be flawed, because so many different
models were tried in order to find the one that gave you the largest
linear coefficient.

		And, consequently, my net conclusion is that more attention needs to
be paid to the possibility of threshold.  Thank you very much.

		DR. HESSEL:  Good morning, Your Honor, folks.  My name is Pat Hessel. 
I'm an epidemiologist.  I was asked to review two papers that OSHA
relied upon in their rulemaking.  The first is a paper by Rosenman et
al.  It was done in -- it was published in 2003.

		What they did was to use the Michigan SENSOR data to estimate the
total number of what they called newly recognized silicosis cases in the
United States.  I think newly recognized is a bit of a misnomer because
what they were really trying to do is estimate the number of
unrecognized silicosis cases.

		The cases were reported to that registry by four sources.  There were
hospitals, physicians, the workers' compensation board, and mortality
data.  When cases were reported to the registry, Dr. Rosenman reviewed
them to determine whether they had an appropriate history of exposure
and whether they either had an x-ray consistent with silicosis or biopsy
evidence of silicosis.

		And then to try to figure out from where he was getting the
information, the three nonfatal sources, so the hospitals, the physician
records, and workers' compensation, from those three sources he used
something called a capture/recapture method to try to figure out, given
the overlaps among those three sources, how many cases were likely to be
in the population unrecognized.

		Capture/recapture is used, has been used to estimate wildlife
populations.  You know, you set a bunch of traps, you catch the rabbits,
you mark the rabbits, and a little while later you set some more traps
and you see how many of the rabbits that you've trapped a second time
were marked.  

		And on the basis of that, you try to figure out how many rabbits there
are all together.  And he used this method to try to figure out how many
unrecognized silicotics there were.

		The purpose was to try to relate the number of deaths, which is a
fairly easy number to track, to the number of silicotics.  So he
tried -- he was trying to calculate that ratio.

		He had 547 silicotics that were non-death certificate, reported to the
registry in 10 years, 547.  Most of those were reported by hospitals,
407 of them, 111 from physicians, and 84, only 84 from the workers'
comp.

		There were 130 deaths in Michigan during that time where silicosis
appeared somewhere on the death certificate.  Michigan only knew about
110 of these, so Rosenman and colleagues excluded the other 20, so they
excluded 15 percent of the deaths reported in the national statistics.

		Dr. Rosenman then reviewed those records and -- of the deaths, and
he excluded another 23 percent where he said although the silicosis was
on the death certificate, he didn't believe silicosis was either present
or had anything to do with the death.  That left 85.

		He then compared the number of living silicotics from the registry,
that 547, to the 85 deaths, and calculated that there were 6.44 living
silicotics for each silicosis death, 6.44 to 1.  And then he used this
capture/recapture methodology to estimate that there were between --
well, sorry, and then he expanded it to all of the United States, and he
estimated that there were between 3,600 and 7,300 new cases per year of
silicosis in the United States.

		Several concerns about this study: one, I think, has to do with the
time period of deaths.  We've seen the charts.  During that time period,
'87 to '96, there was actually about a 38 percent decline in the number
of deaths in the United States where silicosis appeared somewhere on the
death certificate.

		That declined another 55 percent from the midpoint of that '87 to '96
period, until 2006, which was the last information that I had. 
Apparently the information has been updated, and there is continuous
decline.

		Now, obviously, the numbers of cases, of new cases estimated by
Rosenman during that period bear no relevance to the current situation. 
Further, we cannot assume that the quantitative relationship between the
numbers of deaths and the numbers of living silicotics today bears any
resemblance to the estimates that were made in Michigan during that time
period.

		As Dr. Borak pointed out, virtually all of the workers in the
Michigan registry during that time period began their exposures prior to
1970.  Rosenman published something in 1997, a paper where he talked
about the same registry, and he talked about the period '87 through '95.

		Now, the 2003 paper was '87 through '96, so it's only a difference of
one year, but in that -- in the 1997 paper, he showed that more than 95
percent of those workers began their exposure prior to 1970, so prior to
the existing PEL.

		I think we also have to consider the population that makes up the
Michigan registry.  In the Michigan registry -- again, this is from
that 1997 paper, 80 percent of silicotics, or 80 percent of the people
reported to the registry, were from iron and steel foundries.  Of
course, that relates to the car industry in Michigan.

		Now, if we look at the NORMS data for, I believe it's 26 U.S. states
that was published by the CDC in 2005, there were only 6 percent of
deaths where silicosis was somewhere on the death certificate, that came
from workers in iron and steel foundries, so 80 percent in Michigan
versus only 6 percent in the NORMS data.

		Also, if we look at the NORMS data, 22 percent of those deaths, where
silicosis appeared somewhere on the death certificate, were either
miners or quarry workers.  In the Michigan registry, this was only
20 -- sorry, only 2 percent, so 22 percent in the NORMS versus only 2
percent in Michigan.

		Now, in their 1997 paper, Rosenman also presented silica monitoring
data for eight of the Michigan iron and steel foundries that had
produced some of his cases.  They did what they called an initial
inspection and follow-up inspections.

		I believe, on the basis of what was in the paper, the initial
inspections were done in the 1980s -- it sounded like the mid-1980s, so
about 15 years after the implementation of the PEL.

		The average concentration for these eight iron foundries -- this is
unweighted average -- was 1.2 mg/m3 of silica, so we're talking 12
times the PEL, and this was in the 1980s.

		If we look further at this study, there was a single x-ray reader that
was used.  It was only Dr. Rosenman.  He was not blinded to the source
of the x-rays.  He knew that these x-rays were being forwarded to him
because somebody thought that they had silicosis.

		Three-quarters of the people in the registry were smokers.  Now, in my
written comments, I talked about the data relating the reading of
radiographic grounded opacities in relation to smoking history, so I'm
not going to go over that.  But I think there is certainly evidence to
suggest that smoking can enhance the likelihood of a false positive
reading of rounded opacities.

		Rosenman then discussed the assumptions of this capture/recapture
method for his setting, and it didn't appear that any of those
assumptions were met.  One of the assumptions was that there were no
losses or entries during the period of investigation, and he showed that
there was indeed influx into the state of Michigan during that 10-year
time period.

		Another assumption he mentioned was that there would be the same
probability of identification by source, and as I said, the vast
majority of these were identified through hospitals and relatively fewer
from the others.  The other, that ascertainment by sources was
independent, and I think he agreed that that assumption was not met.

		So to conclude the discussion of his paper, virtually all the people
on the Michigan registry were exposed to silica prior to the PEL, the
existing PEL.  The exposures experienced by the people in the Michigan
registry during that 10-year time period were at least an order of
magnitude higher or appeared to be an order of magnitude higher than the
existing PEL.

		The exposure settings of the workers in the Michigan registry were not
typical of workers with silicosis in the United States.  Recall, they
were heavily skewed toward the iron and steel foundry workers.  

		The x-ray readings were susceptible to bias from lack of blinding and
reflected the interpretations of only a single reader.  I'm not being
critical of Dr. Rosenman's reading.  I don't know if he's a good reader
or not.  Nonetheless, it was only his readings that are being relied
upon.

		And then the assumptions of the capture/recapture method did not
appear to be met.  I think it's safe to assume that the results of that
study cannot be used to estimate the current burden of undiagnosed
silicosis in the United States.

		But now if we look at the OSHA document in the Federal Register, if
you look at Page 56384, it's actually in the third column, the first
full paragraph.  And I'll quote here.  "Based on Rosenman et al., 2003,
the Agency estimates that between 2,700 and 5,475 new cases of silicosis
at an ILO rating of 1/0 or higher occur annually at the present PELs as
a result of silica exposures at establishments within OSHA's
jurisdiction."

		There are a number of things I want to talk about in that regard.  I
tried to figure out where those numbers came from, and in this document,
the most recent number they had for the number of deaths was 161.  And
so I plugged the 161 into Rosenman's formulas in his 2003 paper, and I
actually came up with the 2,700 with rounding error, the 2,700 and the
5,475.

		So it would appear that what the OSHA folks have done is to take that
161, apply Rosenman's method, and come up with these numbers of 2,700
and 5,475.  Now, actually in 2006, which was the last number that I had
access to, the number was around 120.

		If we use that number of 120, Rosenman would have decreased that by
1,500 -- or sorry, by 15 percent.  You remember, Michigan didn't know
about some of those that were known by the national.  If he had reviewed
those, again looking at what he did in the period '87 to '96, he would
have eliminated another 23 percent.

		And also, the national numbers include mines, and the CDC paper, 2003
publication, indicated that about 22 percent of the deaths were from
people in mines, and I don't believe OSHA is responsible for that.  So
they would be reduced by another 22 percent.

		When I --

		JUDGE SOLOMON:  I'm sorry.  I have to -- but is there a way that you
can summarize this in the next couple of minutes?  Because we're sort of
at the end of the allotted time.

		DR. HESSEL:  Yes, the joy of going last.  Yes, I will try.  The
numbers I came up with were less than half of the numbers that OSHA came
up with, but that almost isn't relevant.  The workers in the Rosenman
study were exposed before the PEL was put in place.

		At least some of the Rosenman foundry workers had very high exposures
even after the PEL was established.  And so when the statement is made
that these estimated cases are the result of exposures at the present
PEL, it's a statement that simply cannot be defended.

		I was also asked to review another paper.  I have filed comments on
that other paper.  It's just not a terribly informative paper.  But if
you would like to proceed, I will just refer you to my written comments
on that.  Thank you very much.

		MR. CHAJET:  Your Honor, could we have two minutes just to ask
Dr. Hessel to summarize that other paper?  We'll do it in two minutes.

		JUDGE SOLOMON:  Okay.  Now, you have time this afternoon, or you --

		MR. CHAJET:  We do, and we're happy to make up that time later.

		JUDGE SOLOMON:  Okay.

		DR. HESSEL:  In two minutes, sure.  The Goodwin paper tried to, again,
look at the burden of undiagnosed silicosis.  They took a bunch of
people who had died in hospital from respiratory diseases.  As best they
could, they read the x-rays.  Many of the x-rays were very poor x-rays.

		They found some that showed signs consistent with silicosis, some that
showed signs consistent with asbestosis.  The numbers that they came up
with and the percents they came up with are fairly comparable to what
you would see in an unexposed group of that same age, further
complicated by what they call the noise of these terminal diseases on
the x-rays.

		It was an ill-conceived study that really doesn't tell us much.  We
know that there is some silicosis that is undiagnosed.  Their attempt to
actually produce a number was unsuccessful and uninformative.

		JUDGE SOLOMON:  Okay, thank you.  We're going to take five minutes for
a break.  Then we're going to come back, and we'll have about 10 minutes
of questioning from the peer review.  We'll have about 10 minutes from
the public.  Then we'll have some questions from OSHA.  Then if we have
any time left before lunch, we'll come back to the peer review and to
the public.

		Now, from what -- the way I understand it, we'll have some additional
time this afternoon if we have to go back to the peers and to the
public; is that right?

		MR. CHAJET:  Correct.

		JUDGE SOLOMON:  Okay, so we're going to take five right now.  We're
off the record.

		(Off the record.)

		(On the record.)

		JUDGE SOLOMON:  All right, we're back on the record.  At this time
Ms. Lindberg is going to make a motion, or several motions.

		MS. LINDBERG:  I would just like to ask the presenters if they have
materials that we can include in the record, your presentation materials
and your testimony?

		MR. CHAJET:  Allow me to respond please, as counsel.

		MS. LINDBERG:  Sure.

		MR. CHAJET:  Our comments are in the record.  We have a full set of
comments by each of the witnesses.  They're all part of the record
already.  They were submitted to you.  These comments that they made are
in the transcript, and we have no additional documents that reflect
their testimony.  We stand on the oral testimony and the written
comments.

		MS. LINDBERG:  How about the PowerPoint?  Can we put that in the
record, since no one can -- could see it in the record?

		MR. CHAJET:  I think we could put the PowerPoint in the record, sure.

		MS. LINDBERG:  Okay.  Do you have a copy that I can mark?

		MR. CHAJET:  I do not.

		MS. LINDBERG:  May I use this copy I have here that I have not written
on?

		MR. CHAJET:  I will provide a copy for the record after the close of
the hearing.

		MS. LINDBERG:  Okay.  Let me just say that I will reserve Exhibit
Number 6 for Dr. Borak's PowerPoint presentation.

		MR. CHAJET:  Sure.  Actually, you can reserve Exhibit Number 6 for the
entire PowerPoint, which has a few more slides on it.

		MS. LINDBERG:  Sure, that's fine.

		And then I heard we have some vials here that you would like us to
also include in the record.

		MR. CHAJET:  We'll address those at the beginning of next panel.

		MS. LINDBERG:  Okay.  Great, thank you.

		JUDGE SOLOMON:  Okay.  Mr. Perry, you were going to make some
introductions, members of the peer reviewers?

		MR. PERRY:  Well, I introduced them this morning.  I would request
that they be invited to ask questions of the panel ahead of other
participants.

		JUDGE SOLOMON:  Okay.  I have already said that we'll take about 10
minutes to do this.  So if any of you have questions, we're using this
microphone.  You can come forward and stand behind the microphone. 
Please state your name for the record.  Apparently we have the race to
the microphone.

		UNIDENTIFIED SPEAKER:  You probably are --

		DR. CRUMP:  My name is Kenny Crump.  There's been a lot of discussion
this morning of thresholds.  I'd like to ask some questions about those
comments.

		Dr. Borak, in your written comments, you made the statement, "It is
generally impossible to prove a negative."  So you mean by that -- a
threshold is a negative, right?

		DR. BORAK:  No, sir.  I was meaning that it was generally impossible
to prove that there was not something, as a logical premise.  I was not
speaking of the nothing as the threshold.  I was talking about the
impossibility of proving that it doesn't exist.

		DR. CRUMP:  Okay.  So are you saying that -- agreeing that any data,
no matter how extensive, is consistent with no response?  Would you
agree with that?

		DR. BORAK:  I'm sorry.  Could you ask that again, sir?

		DR. CRUMP:  Any data, no matter how extensive, is consistent with some
level of response, in other words, no threshold.

		DR. BORAK:  I must be feeling very dead in my head.

		DR. CRUMP:  Let's take an example.  You have two groups.  One group is
unexposed and one group is exposed.  The unexposed group, there is 0 out
of 100 responders.  The exposed group, they're also 0 out of 100
responders.

		DR. BORAK:  Yes.

		DR. CRUMP:  Can you use that data to reject the hypothesis that there
is no threshold?

		DR. BORAK:  I think you could use the data to suggest that there's no
effect.

		DR. CRUMP:  Can you reject the hypothesis that there is no threshold?

		DR. BORAK:  I'm not sure if it's meaningful to suggest a threshold for
something for which there is no effect.  It's a logical structure that
I'm having difficulty with.

		DR. CRUMP:  Well the scientific method has been described as posing
falsifiable hypotheses and using data to test those hypotheses.  So
declaring an unfalsifiable statement to be true would not be scientific
under that definition of science.

		DR. BORAK:  I didn't think my intention was nearly as profound as your
questions are, sir.  You know, I'm challenged to try to deal with it. 
My intent was to say that -- and perhaps I should look at my comment
and be completely clear.

		JUDGE SOLOMON:  Well, let me just say, we're getting very far afield. 
Had this been a court proceeding and had there been an objection on
whether or not, A, the question was argumentative and, B, whether the
question was answered, there probably isn't an answer.  So you can take
that inference if you want to do that, and ask another question.

		DR. CRUMP:  Okay.  I'll ask another question.  

		DR. BORAK:  Thank you, Your Honor.  

		DR. CRUMP:  Dr. Valberg, you also made a number of comments about
thresholds.  And for example, you say that one time, that Kunkel et al.,
2001, analyzed the responses of rats exposed to high levels of inhaled
crystalline silica, and identified a threshold behavior.  And you said,
i.e., a low L for chronic inflammation.

		I have two questions about that.  Identified a threshold behavior, if
it's impossible to prove a negative, how can you identify something
that's impossible to prove?

		DR. VALBERG:  I guess I take a more practical viewpoint of thresholds.
 And if, in fact, you're exposing animals and you do not get a response,
I mean, I would say that that's a -- certainly is consistent with a
hypothesis that you are exposing them to levels that are below a
threshold, and that you would then have to increase the response to
see -- increase the exposure to see whether a response comes about.

		DR. CRUMP:  So you would consider a low L to be a threshold?

		DR. VALBERG:  I guess, constrained by the biases of how many animals
and what the accuracy of the exposure assessment was, et cetera, that if
you have no response, then that is a no effect level, let alone a no
adverse effect level.

		DR. CRUMP:  Okay.  So you do not distinguish between observed adverse
effect level and an adverse effect level?

		DR. VALBERG:  I guess all of science is based on observation, and so
you are giving me the hypothetical of an experiment that reported no
adverse effects.  And so then, I guess, maybe you would have been --
described that as a no observed adverse effect level, and that's -- I
would agree with that, because anything that we do in science is going
to be ultimately based on observation.

		DR. CRUMP:  Yes.  Would you not agree that the same low L could have
come from a study that did not have a threshold, a dose response that
did not have a threshold?  A small increment of increase that did
not -- but did not have a threshold could have risen to the same low L.

		DR. VALBERG:  It could have given a zero response.  That's correct,
and that's because of the statistics of the situation that, yes, if you
have a low responding -- no response doesn't necessarily mean that it
would always be zero if you had tested an infinite number of animals.

		DR. CRUMP:  Yes.

		JUDGE SOLOMON:  Again, we're pressed time, so maybe --

		DR. CRUMP:  Thank you.

		JUDGE SOLOMON:  -- you have another person in line.  Now, maybe we
can come back to this later.  Is that possible?

		DR. CRUMP:  Yes.

		JUDGE SOLOMON:  Okay.  State your name, please.

		DR. GINSBERG:  Gary Ginsberg.  I'm one of the peer review panelists. 
And there's a couple of questions I have that would help me in
formulating my final comments on these proceedings.  So I think I'll try
to limit these to two or three questions, and maybe I'll just put them
out first and then see, you know, how the responses go.

		So, Dr. Valberg, regarding the Kunkel et al. study that was just
brought up by Dr. Crump, you talk about a threshold for the
inflammatory process being identified.  You didn't use this terminology,
but in the paper they used the terminology of the Mcrit, or the critical
loading of the lung with silica.  

		The specific number is 0.39 mg per gram lung tissue that they identify
in the paper, and you talk about that as a threshold for inflammation. 
But my read of that paper is a little different, specifically in that
where they talk about the no L's and low L's in Table 2, they clearly
show that they did not find a no effect level, that the lowest dose was,
in fact, an effect level for the increase in PMNs.

		And they talk about that specifically in the paper as there was no
threshold for that effect.  And then if you look at their graphs, the
PMN graph as well as their graph on alveolar epithelial hypertrophy and
hyperplasia also shows no evidence of a threshold.

		So I think that it's worth, you know, being careful about what, in
that paper, did show a threshold -- you know, the fibrotic, clearly,
the initial instigating events, clearly no threshold, as well as also
their analysis of the variability that might be inherent in their data,
where they look at the Mcrit, the 95 percent lower confidence limit on
the Mcrit not being 0.39 anymore but being 0.0049, which means a much
lower inhalation dose.

		I think it was instead of 0.036 mg/m3, it's 0.007 mg/m3, that would be
the Mcrit for the most sensitive individual.

		And one thing that has been missing, I feel, both from the OSHA
analysis as well as from the comments I've heard so far, is any
discussion of variability in the population that might, on the one hand,
tend to linearalize or obscure -- I'll call it obscure a threshold, not
just error -- measurement errors, as Dr. Long so rightly talked about,
but there's other factors in terms of population variability, as is
brought out in the Kunkel paper, that that Mcrit value could vary.

		So I'm sorry for the long question, but that's one, if you want to
respond back to that.  The other, I wanted to ask Dr. Long, so how much
measurement error do you need?  If you have a threshold, let's say,
that -- a true threshold, if such a thing is possible, of 0.05 mg/m3,
and you have a bunch of measurement error, how big does that measurement
error need to be for you to, say, increase that to 0.15, a threefold
shift?  How does that work?  I'm not a statistician.

		And then the third piece of my triumvirate questioning is that it
appears to me, that -- self-evident, as Dr. Borak talked about things
that seem to just fall out of the data, that there is a number of
studies where the highest dosed groups have a flattening of the response
either to cancer or to silicosis.

		I'm talking about the granite, the Graham study, the diatomaceous
earth studies, the most recent Liu Chinese studies, and a number of
others.  The high dose groups don't have the same kind of dose response.
 

		And those high exposed people would have the highest measurement error
because they were the ones that were in the bad old days when there was
no PEL, et cetera, et cetera, and the measurements would be -- so it
seems logical to me to think about dropping off the weight put on those
older, higher exposed groups because you've got the biggest errors.

		As you rightly point out, measurement error is important to consider. 
And you've got the best data on those workers that have the more modern
measurements.  So those are my three parts.

		JUDGE SOLOMON:  Okay, let's start with Question Number 2 to Dr. Long,
because that's probably the easiest way to marshal the questions. 
Dr. Long?

		DR. LONG:  Yes.  I can try to answer that.  I'm not sure that I have a
definitive answer.  I'm not a statistician either, I'm an exposure
assessor, but I can tell you that that's not an easy question to answer.
 It depends.  I mean, there are a number of factors that influence --

		DR. GINSBERG:  Variability in a population.

		DR. LONG:  Yes.  There are a number of factors that influence what
impact exposure measurement error can have, not only the size of the
exposure measurement error but also the error structure, the type of
model that is being used, you know, other sources of variability. 
It -- I guess the answer is, it depends.

		JUDGE SOLOMON:  Okay, so Number 3, Question Number 3 goes to validity;
is that right?  Is that what you were really asking about?

		DR. GINSBERG:  I'm just looking for how much measurement error could
affect the threshold.  And do you have a study I could go home and read
that talks specific -- quantitatively about that?  I mean, there's
a -- I've seen a lot.  I've read the Romberg and I've read the 2011
paper.  There's not a lot that's quantitative about that.

		DR. LONG:  There are some quantitative examples in the Romberg paper. 
There's a -- I think it's a study by Carolyn Cuckanoff (ph.) that kind
of looks at different sizes of exposure measurement error and how that
influences the shape of the exposure response curve.  So I think that is
a quantitative example.

		But, you know, again, I mean, the answer is, it really depends on a
number of factors.  So it -- I think that's a difficult question to
kind of pinpoint how much error can cause a threshold to be obscured or
underestimated.

		DR. GINSBERG:  So there'll be --

		JUDGE SOLOMON:  Okay, so in order to marshal all this -- okay, this
is like a compound, loaded series of questions.  Why don't we just give
the panel a chance to answer?  Dr. Long has already spoken.  Does
anybody else want to address any of these issues?

		DR. VALBERG:  I'll just respond.  I mean, I really appreciate your
comments on the Kempel et al. paper.  I think you're correct.  I mean,
there is a lot of nuances that need to be considered, and I agree with
your comments.

		I think the issue, in terms of actual affect, of course, then boils
down to discriminating physiological responses which occur in response
to inhaled particles versus ones that are persistent and are
irreversible and become wholly adverse and lead to disease, but I think
you're correct.  I mean, that -- I didn't mean to simplify it.  I had
very limited time.

		JUDGE SOLOMON:  Anybody else?  Okay.  Thank you very much,
Dr. Ginsberg.  We'll come back to some of this, this afternoon.  

		Now, are there any of the members of the public who wish to come
forward?  If you do, we'll take about 10 minutes of questioning, and
then we'll go to the OSHA people.  State your names, please.

		DR. GOLDSMITH:  Yes, thank you, Your Honor.  My name is David
Goldsmith.

		JUDGE SOLOMON:  Can you spell your name where the Court Reporter --

		DR. GOLDSMITH:  G-o-l-d-s-m-i-t-h.  I'm going to speak tomorrow, but
I'm here today, and I'm seriously interested in --

		JUDGE SOLOMON:  And in what capacity are you here for?

		DR. GOLDSMITH:  I'm here as a representative of George Washington
University in Georgetown, but I'm not paid by anyone.

		I have just an observation and a question for the panel.  Back behind
you all is this graph of CDC decline in silicosis mortality, and I would
tend to suggest that there is certainly a reason for something positive
to be said about this.  

		And the positive things need to be said about efforts and beneficial
effects put forward by industry to lower the exposure to silica dust,
and that has led in part to this decline in mortality rates, in addition
to which it's also clear that OSHA's raised people's awareness about
this.  And that's another possible explanation why there is this clear
decline in silicosis mortality.

		But my question is, is this the right thing to look at?  Shouldn't we
be looking at all of the silica-related disease, not just silicosis? 
Shouldn't we be looking at lung cancer, autoimmune disease, kidney
disease, and tuberculosis, which has fallen dramatically at the
beginning of this time frame, as indications of an effective policy put
in place by industry, and a regulation that's been sort of backed up by
OSHA in terms of telling both workers and the community that action is
needed?

		And I think that this is something that needs to be put into some
context, because as I see it, cases of silicosis that were at the
beginning of this period, these individuals are still there, they're
dying of other diseases.  And as -- other diseases as simple as falls
and automobile crashes, heart disease, what have you.

		And so it's somewhat narrowing, in terms of looking at this whole
question about the role of industry and the role of OSHA, to look solely
at silicosis mortality.  And I just would be pleased to hear what the
panel has to say about that.

		DR. BORAK:  Let me take a first cut.  I agree with Dr. Goldsmith.  I
think that the slide that you are seeing is a tribute to an important
public health success.  Whether it's complete or not is a separate
issue.  It reflects a substantial decrease in deaths related to a
preventable cause.  Whether those exposures occurred before or after
OSHA should not diminish the magnitude of the success.  That's the first
thing.

		I spoke of this curve, essentially to speak to whether this
information informed the evaluation of whether the current PEL was
adequately or was not adequately protective.  That's a different issue
than you bring up, and so let me separate these things, because your
question was multi-part.  

		First, this is a sign of a success.  It has been a dramatic decrease
in mortality, and additionally I showed some slides which indicated a
dramatic decline in the exposure levels that occurred in this country,
but also in China.  And I presume we could see the same kind of dramatic
decline elsewhere.

		It's clear that the decline was a consequence of industrial activity,
that is say, activity by corporations and companies and others who were
responsible, and it is also attributable to the implementation and
adoption of various regulations and rules.

		And it's clear that in China, the decline occurred when rules came
into effect, from horrific levels.  And I showed you data that come from
the OSHA IMIS database indicating 50 to 90 or 95 percent declines in
ambient silica levels in workplaces, or silicate workplaces, as
discussed by OSHA scientists, and that's a tribute to both industry and
OSHA working together.  And I think that this has been a wonderful
success.  So that's my contribution to answering your question.

		DR. GOLDSMITH:  And don't you also think that we should be looking at
a broader set of diseases than simply looking at silicosis mortality?

		DR. BORAK:  Ah, yes, I think we are looking at a wide array of
diseases.  This happens to be a clearly documented and dramatic
statement regarding one of those most important diseases, one of which
is a signature disease.

		I think that the question is not that -- I, certainly, as a
physician, do not regard any of these diseases as unimportant, either to
people or to the public health.  My concern had to do with whether 100
ppm PEL was or wasn't protective, and whether there were data, okay. 
But I certainly would not tell you that these other diseases are
unimportant.  That would be a terrible thing for me to say.

		JUDGE SOLOMON:  Okay.  We're limited with time.  We have one other
member of the public who wants to ask some questions.

		DR. PARK:  Robert Park, NIOSH, from Cincinnati.  There's been a lot of
interest in what's going on at the low end of the exposure response
curve with a special focus on threshold.  I haven't heard any
speculation about the possibility of an increase in the exposure
response at the low end of the curve. 

		Maybe something's being up-regulated and there's a threshold.  Maybe
at the very low level, risk is higher than you would predict from a
linear overall exposure response.  Why are we not hearing of that
speculation?

		Secondly, a lot of talk about measurement error.  If there's no
underlying association, that is not going to generate statistically
significant associations.  It's going to generate random estimates of
exposure response.  Half of them will be negative, half of them will be
positive.  If it's a big study, they'll be really small.  If there's no
association, we don't have to worry about measurement error.

		On the question of latency for silicosis, that's just an artifact of
how much exposure the person has had.  If they're at low exposure, it's
going to take longer.  You're going to have a 35-year latency.  If they
have high exposure, if it's high enough it could be one year.  So
defining a threshold based on when silicosis happens is an artifact of
clinical detectability.

		JUDGE SOLOMON:  Let me put a question mark at the end of that. 
And -- is that a question?  Or are you just --

		DR. PARK:  Just one more sentence.

		JUDGE SOLOMON:  Okay.

		MR. CHAJET:  I'm going to object --

		DR. PARK:  Therefore, how can one assign a threshold for lung cancer
to this magical threshold, which depends on sort of the history of
average exposure and clinical detectability?

		JUDGE SOLOMON:  Okay.  What's the nature of the objection?

		MR. CHAJET:  Well, I hadn't had a question for the first three
minutes, so if --

		JUDGE SOLOMON:  Well, this is an example of the loaded question, but I
believe that there -- that as long as somebody can answer --

		DR. PARK:  Why isn't there a positive effect at low doses?  That's a
question.  Why aren't you speculating about that?  Why is measurement
error important --

		JUDGE SOLOMON:  Why don't you give them a chance to answer?  Okay,
does anybody want to take that one?

		DR. VALBERG:  Yes, thank you for that question.  And the speculation
about what is the actual nature at very low doses is a good one.  I
think that it is interesting, you know, how would the organism begin to
respond from very low exposures to minute exposures and so forth?  

		And it could very well be that there's all sorts of things that go on.
 You may have heard of the word "hormesis," that in fact maybe zero
exposure is actually not as good as some level of exposure.

		DR. LONG:  I wouldn't go that far, but yes.

		DR. VALBERG:  But yes, I'm just saying.  But yes, the shape there is a
little bit unusual.  And I think that part of the reason that I didn't
include it is because the whole idea is a little bit speculative and I
don't -- I'm not meaning to say that we should dismiss it out of hand. 
We should keep it in mind, and if we find evidence that can be helpful
in setting standards, we should use it.

		JUDGE SOLOMON:  Okay.  I think at this point we'd better go to the
OSHA questioning.  Thank you very much, Dr. Park.  

		Mr. Perry?

		MR. PERRY:  Thank you, Your Honor.  The first panel member to question
the panel will be Steve Schayer from our Office of Physical Hazards.

		JUDGE SOLOMON:  Technical difficulties beyond our control probably
require another microphone.

		(Off microphone discussion.)

		MR. SCHAYER:  Well good morning, and thank you, Bill.  So I have a
question for the panel.  In your written comments to OSHA, you mention
several examples of potentially large sources of exposure measurement
error not considered by OSHA.  

		These included between measurement variability, between sampler
variability, between laboratory variability, between sample duration
variability, between sample location variability, between sample type
variability, and error from particle potency variation.

		So I was wondering if you could kind of clarify on what data would
need to be available to address these sources of error.

		DR. LONG:  Great.  Actually, at this afternoon's session we'll be
talking -- we'll be addressing several of these sources of error, in
particular sampling errors, analytical errors, pump errors, so I think
that session will address, you know, provide information on some of the
data sources needed to address them.

		I think, regarding some of the other sources of error, such as between
sample, I mean, looking at, let's say samples that are collected as part
of compliances purposes versus random exposure survey -- data from
random exposure surveys, I mean, those data are available.  I would
think you can characterize, you know, the size of exposure measurement
error.

		So I think, regarding several of the sources of exposure measurement
error, there are data available.  There are data from area samples
versus personal exposure samples.  You can characterize the difference
between them.

		MR. SCHAYER:  Okay, thank you.  Another question that we have as well
is could you please identify some studies that you believe do acceptably
account for the impact of exposure measurement error on exposure
response associations for respirable crystalline silica?

		DR. LONG:  You know, off the top of my head, I can't think of any
studies that have done that.  I can go back and think about that
question and see if I can identify appropriate studies, though.

		MR. SCHAYER:  Okay, thank you.  What about studies of other
occupational epidemiology cohorts?  Can you give us some examples of
other types of occupational studies that have addressed these sources of
measurement error that you've brought up today?

		DR. LONG:  Again, I can't think of any specific studies.  I can go
back and look into that question.  I believe that exposure measurement
error has been addressed in studies of asbestos-exposed workers.  I
can't think of any specific studies off the top of my head, but I can
get back to you with information.

		MR. SCHAYER:  Okay, thank you.  And one other question that I have,
for the studies that Dr. Valberg cited, that showed evidence of a
threshold, these would be Graham, 2001, 2004; Pukkala et al., 2005;
Rushton, 2007; See (ph.), 2007; and Morfeld et al., 2013, we were
wondering if you believe that these studies actually address the sources
of exposure measurement error that you said the OSHA studies had not
addressed.

		DR. VALBERG:  I appreciate that question.  On those particular
studies, I mean, my -- I did not look at exactly how they treated
exposure measurement error, but I think to the extent that they were
able to discover a threshold, they essentially did not have a surfeit of
noise in their exposure data, and that the threshold could be revealed.

		I think one of the key ingredients in looking for thresholds is that
you have to begin with a model that could allow for that possibility in
order to be able to detect it, and then look for some sort of likelihood
as to which model fitting is going to be the better one.

		MR. SCHAYER:  Okay, thank you.  And one other question that I have,
too.  So you've noted today in your comments and also in your written
comments as well as today's testimony that exposure measurement error
can bias regression coefficients, modify the shape of exposure response
curves, and obscure the presence of a threshold.

		So isn't it also the case that exposure measurement error can bias
results towards the null?

		DR. LONG:  Yes, that is true.  It -- yes, as I discussed in response
to Dr. Ginsberg's question, the impact of exposure measurement error
does depend on a number of different factors, and it is true, it can
bias either high or low.

		It can bias towards the null.  It can be a source of positive bias. 
Yes, there are examples of that.

		MR. SCHAYER:  Okay.  Thank you very much.  That concludes my
questions.

		JUDGE SOLOMON:  Mr. Perry?

		MR. PERRY:  Next up will be Tiffany DeFoe.

		MS. DeFOE:  Hi.  I think a number of my questions have been pretty
well addressed by some of the previous questioners and by the panel, but
just a couple additional.

		I wanted to follow up on a question that Dr. Ginsberg, I think,
raised.  Dr. Valberg, would you -- well, first, would you quickly just
step through the beginning of the disease process and how it relates to
the possible existence of a threshold effect in silicosis?

		DR. VALBERG:  Yes.  I'd like -- I'd be glad to do that.  I think, one
of the things that's important is that whenever we're -- whenever the
lungs are presented with a challenge of inhaled gases, vapors,
particles, and so forth, the lung defense systems can kick in, if there
in fact is some sort of adverse reaction with the lung tissues.

		And you can have a enhancement of the lung defense mechanisms that
helps you deal with the challenges of everyday particle inhalation that,
in fact, is a psychological response.  And even though you may be able
to detect a difference between exposed lungs versus unexposed lungs,
that difference may or may not necessarily be adverse.

		And so it's the persistent lung inflammation that seems to cause
problems in the long term.  And if you get that, either by virtue of an
overload of the lung or by virtue of particles that are particularly
toxic in their reaction with cells, then that can go on to be
persistent, irreversible, and lead to disease.

		So what I was picking up, in terms of people who have examined this
reaction between silica and the lung, is that the primary step is
inflammation, and that if that gets out of control due to an overload
condition, then you can have lung disease.

		MS. DeFOE:  So the clearance mechanisms are important to preventing
that sort of overload?  Are there other mechanisms?

		DR. VALBERG:  Yes, absolutely.  No, I think that the whole response of
the lung does involve, you know, dissolution, clearance, the removal of
the material from the lung.  And particularly for particles that are
insoluble, then it is of course macrophage clearance, mucociliary
clearance, perhaps clearance to the lymph nodes as well.

		MS. DeFOE:  And it sounded like you -- is it the case that you would
agree with Dr. Ginsberg's statement that these processes vary quite a
bit from individual to individual?

		DR. VALBERG:  Yes.  And that's one thing I meant to respond to
Dr. Ginsberg's question about individual-to-individual variability. 
That does indeed exist, and that's another factor that might tend to
obscure a threshold; something that might be steep for a very homogenous
population, if there's variability in a population, then that threshold
can be blunted, if you will.

		And Dr. Ginsberg served on the panel that developed the National
Academy of Sciences' so-called Silver Book, which had to do with
responses to -- on non-carcinogenic effects, that individual
variability is indeed very important.

		The only factor to keep in mind, however, is what is the degree of
individual variability?  That is to say, you need some information on
how much people would vary.  It's not infinite.  I mean, all of us
manage to live in the face of lots of individual challenges.  People
manage to put out drugs into the drug store that we all take, and
seemingly they are in a range that do not, in fact, kill people and are
effective for most people as well.

		So yes, absolutely, individual variability is an important parameter.

		MS. DeFOE:  Are you aware of good studies on the nature of and extent
of the inter-individual variability on these parameters?

		DR. VALBERG:  There are studies on lung responses to particles and
individual variability.  I don't -- I can't, you know, cite them to you
right now, but in fact that's a fairly well-studied field, yes.

		MS. DeFOE:  Thank you.  And then just a clarifying question to
Dr. Borak, you stated earlier that OSHA relied on a hypothetical risk
assessment rather than the empirical data itself.  That was a phrase
that I picked up from your earlier statement.  And I wanted to clarify,
when you said the empirical itself, which data were you referring to?

		DR. BORAK:  I was talking in particular, in the -- what OSHA refers
to as the surveillance data.  And when I refer to the hypothetical, what
I am talking about is the modeled data largely driven by high exposures
in which the low exposure extrapolations are made.  And I think that
those extrapolations have a certain degree of uncertainty and are
thereby hypothetical.

		MS. DeFOE:  So then if you're suggesting that OSHA should rely on the
empirical data itself and you're talking about surveillance data, are
there empirical data available on the exposures of these workers, or on
their use of respiratory protection?

		DR. BORAK:  No.  I think that I agree with OSHA in its original
statement, which was that the surveillance data are insufficient.

		MS. DeFOE:  Okay, so you didn't actually mean that we should rely on
the surveillance data?

		DR. BORAK:  No, no, no.  To the contrary.

		MS. DeFOE:  Okay, thank you.

		MR. CHAJET:  I have just a couple of questions.  Dr. Borak, I wanted
to ask about your recommendation for changing the definition of
respirable crystalline silica in our proposed rule.  In particular, you
say that the definition should be defined by particle size.

		So are you saying the definition should refer to all particles less
than, say, 10 microns in diameter or less?

		DR. BORAK:  It's -- my objection has several parts.  One of them is,
as you will hear this afternoon, and I don't want to take away that
discussion, each of the individual samplers which has been proposed or
might be proposed has a different profile in terms of the sampling
efficiency at various sizes, aerodynamic sizes of particles.

		And so one of the problems is that some samplers will under-sample
very small particles and over-sample larger particles.  And others work
in the other fashion.  This has several implications, not only in terms
of how much appears to be in the air, but also from a biological
standpoint.  

		If you believe, at least as I do, that it's perhaps the surface area
of the particles and not their mass which is a major driver of the
toxicology of silica, then those samplers which under-sample small
particles and over-sample large particles will give very misleading
information as to the toxicity of the dust and at the same time will
lead to incorrect assumptions as to whether people are or are not above
or below allowable levels.

		And I think that is one of the things OSHA needs to consider, which is
that there needs to be a absence of such ambiguity, and that one of the
problems has to do with the variability of different samplers, resulting
in variability in terms of the actual data and their implicit or implied
toxigenicity.

		The second has to do with a problem which includes the fact that there
are sometimes errors in which large particles are inadvertently included
in samples, which give misleading readings of mass and so forth.  And I
think it would be very useful if this was not left to a question of
sampler performance, but rather to a standard which was size based.

		MR. CHAJET:  Okay.  Let's see, I think what I was trying to ask,
maybe, was a little bit different.  First, we all understand that these
respirable dust samplers will collect particles of different sizes with
varying degrees of efficiency.

		The larger particles will be collected less efficiently than the
smaller particles.  Do you agree, as a general matter with that?

		DR. BORAK:  I think so.  I'd have to think twice, but for the moment,
yes.

		MR. CHAJET:  Well, I'm trying to figure out, if we define respirable
crystalline silica as being the particles that are less than 10 microns,
meaning the mass of all of these particles, and we express whatever
exposure limit we end up with as a mass of all of these particles, is
there a sampler available that can measure that, if we define respirable
crystalline silica this way?

		DR. BORAK:  I have to admit that I don't know the answer, but this may
or may not be a subject for this afternoon's discussion.

		MR. CHAJET:  I may hold it till then; thank you.  One other thing.  I
can't remember now who it was, but maybe it was you, Dr. Borak, quoting
IARC 97 with respect to what IARC concluded, that the inflammatory
reaction was a likely mechanism for the induction of lung cancer?

		DR. BORAK:  That's correct.

		MR. CHAJET:  Okay.  Have you seen their updated assessment in 2012?

		DR. BORAK:  Yes.  That was quoted just below, on the same slide.

		MR. CHAJET:  Okay.  Then it should be on your slide that what you say
is true, but then they go on to say, "Although reactive oxygen species
can be directly generated by crystalline silica polymorphs themselves,
and can be taken up by epithelial cells, for this reason, a direct
effect on lung epithelial cells cannot be excluded."

		DR. BORAK:  Yes, correct.  I think I actually quoted that last
statement, and that may have been --

		MR. CHAJET:  I don't remember that.

		DR. BORAK:  I'm sorry.  Was that a question?

		MR. CHAJET:  No, no.  I just couldn't remember you quoting that
particular statement.

		DR. BORAK:  Oh, oh.  It's at least in my written comments and is tied,
in part, to the question that Dr. Crump asked me earlier.

		MR. CHAJET:  Okay.  All right.  Let's see.  The only -- just one
other thing.  In your -- and this is the group now, that you all
mentioned a lot of published studies, and Dr. Long, I think, in
response to a question from Mr. Schayer, you talked about perhaps other
occupational epi studies that would be helpful to us.

		I just wanted to make sure.  Have all of these been submitted to the
record already?  Or will you be able to submit anything that you've
talked about in your oral testimony today to the record?

		(Off the microphone discussion.)

		MR. CHAJET:  Ms. Lindberg has a few questions for the panel.

		MS. LINDBERG:  Dr. Borak, I'm going to quote from your comment that
you submitted to the record, Pages 9 to 10.

		DR. BORAK:  I'm sorry, page?

		MS. LINDBERG:  Pages 9 to 10.  And the quote is, "It should be obvious
that because of the widespread noncompliance in silica overexposures
documented by OSHA, and because of frequent noncompliance with
respiratory protection, there is little or no empirical basis for OSHA
to attribute ongoing reports of new silicosis cases to exposures at or
about the current PEL."

		And perhaps it's because I'm an attorney, but it doesn't seem obvious
to me that this proposition is true.  So I guess I'm wondering, have you
been able to or attempted to demonstrate that these new silicosis cases
were linked to silica exposures above the current PELs rather than below
them?

		DR. BORAK:  I have not been able to relate individual cases, nor has
OSHA been able to relate individual cases to levels at or below the 100
level.  I think the weight of evidence is rather compelling and -- at
least rather compelling to me.

		One of them is demonstrated in the slide which is up here, which shows
a dramatic decline in silicosis, despite the persistence of levels above
100 µg.  And there is a lot of data showing that the exceedances and
noncompliance with the current PEL seem to be most commonly found in
those industries that are associated with the highest rates of
silicosis.

		So I make that inference, and it seems to me clear, but perhaps it is
not clear.  It may not be obvious to all.

		MS. LINDBERG:  Just so long as I'm not missing something.

		DR. BORAK:  No, no, no, no.  I think it's a very sharp question, and I
appreciate your asking.

		MS. LINDBERG:	 And then one for Dr. Hessel.  Again, going back to
your comments that you submitted to the record, on Page 6 you described
a case involving thousands of silicosis lawsuits, a multi-district
litigation.

		And you state that the judge found that the cases of alleged silicosis
were mostly fabricated for purposes of the litigation.  You didn't cite
the case, but I went back and I found it and looked at it.  And as far
as I can tell, those cases involved in that litigation were from
Mississippi, Kentucky, Texas, and Missouri.  But that was cited in a
section of your comments on silicosis in Michigan.  So I'm wondering if
there was a link between that litigation and silicosis cases that were
discussed in the litigation.

		DR. HESSEL:  Yes.  Thanks for your question.  I thought you'd
forgotten I was over here.  The point I was making, I don't know if any
of those screening programs had taken place in Michigan.  I know that
the similar kind of bogus asbestos screening -- or asbestosis screening
programs had taken place in Michigan, but I did not know if the
silicosis screening had done, had happened there as well.

		If they had happened -- see, what I was trying to figure out is, we
had so many of these cases that were reported by hospitals.  Now, a
hospital can generate a silicosis on the record by a diagnosis that
would occur in the hospital, or simply by taking a history, a medical
history on admission, and if somebody on admission says, well, you know,
yes, I've got high blood pressure, oh, and I've got silicosis because,
you know, those lawyers told me so.  

		So what I'm -- what I was concerned about is that if those kind of
screenings had happened in that jurisdiction, or in Michigan, then
people would be walking around with a diagnosis of silicosis, and they
might present that when they came to the hospital, and the hospital
might just have a more efficient mechanism for referring to the registry
than some of the other situations.

		MS. LINDBERG:  Yes.  I understand your concern.  I was just wondering
if you had any link or any evidence that you could submit that would
actually document that that kind of screening was happening in Michigan
for silicosis rather than asbestosis.

		DR. KESSEL:  Right.  No, I looked hard.

		MS. LINDBERG:  -- the number of cases.

		DR. KESSEL:  Yes, I looked hard, and I couldn't find any.  But that
doesn't mean it didn't happen.  I just don't know that it happened.

		MS. LINDBERG:  All right, thank you.

		MR. CHAJET:  Let me just interject, for the record, we'll try to find
some of that information.  But clearly we had 30,000 or so cases in
front of Judge Jack, and we quote her in our comments where she said,
"This appears to be a false epidemic, or a phantom epidemic."  

		Those cases infest your data as well, because when somebody says they
have silicosis, they go into your database.  And that's 30,000 of them
in one case.  So this is not a phantom problem; it's a phantom epidemic,
as Judge Jack said.

		JUDGE SOLOMON:  That's argument and not testimony, by the way.  But
we're at 12 o'clock, so how much more do we have?

		(Off microphone discussion.)

		MS. LINDBERG:  Just one more for you, Dr. Hessel.  OSHA cites one of
your studies on Page 56299 of the proposal for the proposition that once
someone has been diagnosed with silicosis, there is a chance the disease
will get progressively worse, even in the absence of silica exposure. 
It looks like it was a study from 1988.  I'm just wondering, do you
think that's a correct interpretation of that study?

		DR. HESSEL:  I think, in the population we were dealing with -- these
are underground mine workers in South Africa exposed to some fairly high
levels, and the majority of the people who developed silicosis did have
progression.

		We found that when we looked at the difference between progression in
relation to exposure before and after onset, exposure after onset was
actually important.  So yes, I believe the data speak for themselves.  I
think we found that in most cases there was progression.

		MS. LINDBERG:  Thank you.

		JUDGE SOLOMON:  Anything else?

		MR. CHAJET:  Just one more.  I'd like to ask Dr. Borak one question
as a follow-up to a response of one of Ms. Lindberg's questions.

		In talking about this graph that's up here again, of the decline in
silicosis mortality over the years -- and I think you were talking
about thresholds, do you regard this, these data, as evidence of a
threshold for silicosis at or above OSHA's current PEL?

		DR. BORAK:  I think that Dr. Goldsmith linked this and -- a
question.  I thought it was Dr. Goldsmith, but somebody asked regarding
this graph and thresholds.  I see no correlation between this graph and
a threshold.

		What I was trying to talk about was the chronology.  I'm sorry to be
blowing into the speaker.  I was simply speaking of the chronology and
in particular the concerns about latency and survivalship and the
relevance of these data to evaluating the effectiveness of the PEL.  I
think they are a tribute to both industry and government intervention.

		MR. CHAJET:  Okay.  Thank you for clarifying that.  That's all, Your
Honor.

		JUDGE SOLOMON:  Okay.  So we're going to come back at 1 o'clock, when
we'll have the Chemistry Council.  So the hearing is continued to 1
o'clock.

		(Whereupon, at 12:03 p.m., a lunch recess was taken.)

A F T E R N O O N   S E S S I O N

(1:00 p.m.)

		JUDGE SOLOMON:  All right, let's go back on the record.  We have
Dr. Cox.  You want to make the introduction?  State your name.

		MR. KING:  My name's Neil King.  I am counsel for the American
Chemistry Council's Crystalline Silica Panel, and I'm just here as kind
of a prop for Dr. Louis Anthony (Tony) Cox, who will be making a
presentation this afternoon.  

		DR. COX:  Great.

		JUDGE SOLOMON:  Go Ahead.

		DR. COX:  Okay, so I am Tony Cox.  I'm the President of Cox
Associates, a risk analysis and quantitative modeling firm based in
Denver.  I'm a member of the National Academy of Engineering; editor in
chief of Risk Analysis, an international journal; author of the book,
Improving Risk Analysis; and a fair amount of my life and passion are
devoted to how we can, indeed, do that.

		The American Chemistry Council asked me to critically review the
assumptions and models used by OSHA, and the sources on which OSHA has
relied in this matter, and I've done that.  And I've submitted a fairly
extensive piece of written testimony, back on February 11.

		What I want to do this afternoon is to summarize what I see as being
some high points of that, and to invite, of course, questions and
exploration.

		So let me begin with a preview of what we're going to cover in the
next half hour, a little bit more.  I want to look at what are the
components of a good quantitative risk assessment, as I understand them,
and then say, well, how good is OSHA's risk assessment compared to those
criteria, or by those criteria?

		And then I want to give you my assessment of the scope of the
preliminary QRA that OSHA has.  So is it answering the right questions? 
Is it tackling the right questions?  And also the methods, is it
tackling those questions in correct ways?  So I'll tell you what I think
about that, and that will include touching on some of the topics that
those of you who've been here for a while have already heard.

		So I, too, will comment briefly on some biases, the threshold issue,
how well uncertainties have been characterized, and I especially want to
underscore the topic of causality.  So those -- that's the meat in the
sandwich.  That's the key part of what I want to cover.

		And then I'll switch to conclusions and recommendations.  My
understanding of at least part of the purpose of this hearing is to say
how good is the current preliminary QRA?  Out of that flow
recommendations, saying how might it be done better?  And so I'll close
by touching on those.

		Let me start with what I take to be the goals for a good quantitative
risk assessment, and here I'm drawing heavily on the NRC volume, Science
and Decisions, which emphasizes that a good quantitative risk assessment
is, from the outset, designed to support better risk management
decisions.

		So here, the risk management decision of interest is should we
change -- lower the standard for crystalline silica in order to protect
worker health?  Is that a desirable decision, or will it not produce the
effects that we want it to produce?

		So we want to support improved risk management decisions by revealing
the probable consequences of different choices.  If we do this, what are
the likely consequences?

		And at the same time, we're all imperfect human beings.  We need to
acknowledge there is some uncertainty left.  So if we think we know what
will happen if we do or do not change the current standard, how sure are
we that that will happen?  And an honest characterization of that
uncertainty is important to guiding decisions.

		My approach to reviewing the preliminary QRA was to ask, would this
document -- should this document convince an open-minded skeptic?  So
here I mean a skeptic who starts off saying, well, I don't know whether
this is right or not, but let's walk through and see what we think.  So
would this convince an open-minded skeptic knowledge about risk
assessment?

		And although the NRC suggests quite an elaborate process for
quantitative risk assessment, the three things that I want to focus on
are, first, decisions, on the right hand side, pulling the risk
assessment.  Are we going to be able to reduce uncertainty about the
likely consequences of different decisions enough to be useful?  That's
the sort of key point.

		And on the left hand side, have we formulated the right questions? 
Have we scoped the risk assessment so that it answers those
decision-relevant questions?  

		And in the middle, we have the familiar building blocks of
quantitative risk assessment and risk characterization and uncertainty
characterization and so forth, but with the purpose of supporting better
decisions.  So that's the framework that I'll be following.

		And for those of you who were planning to sleep in the post-lunch
period, let me just state my main conclusion, that I think my best
assessment is that the Preliminary Quantitative Risk Assessment is not
ready for use.  It's not ready for prime time, in some sense.  And
here's why.  And then I want to develop these points over the next half
hour.

		The answers and estimates -- risk estimates that the preliminary QRA
delivers are not trustworthy.  You shouldn't bet money on them.  They
might not be right.  They might not be wrong, but there's -- but
they're not trustworthy.

		Specifically, there are what I take to be important statistical errors
in modeling and inference.  Now, when anybody -- when a statistician
accuses anybody else of making an error, you're entitled to know, error
with respect to what standard?  What's the gold standard that entitles
you to say this is wrong or this is right?

		So I'll be referring to principles such as, if there is substantial
measurement error in exposure estimates, then that measurement error
should be modeled, that I think are uncontroversial.  I think that
people with any preference for the outcome -- yes, we should reduce the
standard, or no, we shouldn't -- would agree or should agree to these
principles.  They're widely accepted in statistics.

		I will touch upon numerous biases that I think have not been
adequately controlled.  And these biases lead to false positives.  It's
a natural question to say, well, couldn't they lead also to false
negatives, that is, failing to find something that's there?

		But for reasons that I'll elaborate on, mainly having to do with the
fact that uncertainties are ignored which artificially contract
confidence bands, the only effect of that is to increase false
positives.  It never increases false negatives.

		I'll talk about exposure measurement error.  And I heard that being
discussed before lunch.  I think it's an important topic.  But I'll talk
about it briefly because it's already been touched on.

		I do want to look at the possibility of thresholds, and I find that
the question of is there a threshold and how do we model exposure
measurement error are closely related.

		Okay, and then I want to talk about how uncertainties should be
characterized to give policymakers an honest understanding, an accurate
understanding of where the reigning uncertainties lie.  I think that has
not yet been well done, not only because exposure measurement error has
not been well characterized, but because modeling uncertainties have
been ignored, by and large.

		And then, thirdly, I'm going to make a case that what we really care
about here is a cause and effect relationship.  If we reduce exposure,
will that cause harm to workers to go down?  

		And that crucial question, which is the one that ought to be of
primary interest to policymakers and risk managers, has not been
addressed at all in the Preliminary Quantitative Risk Assessment, except
for in a couple of passages where OSHA says, we think this is causal. 
But they provide no information about why they think that, no rational
basis for following or refuting their reasoning.

		Almost -- it's almost the case that there is no discussion of
causation other than such opinions, and it is the case that no formal
causal analysis or causal modeling has been presented.  And this, to me,
is crucial.  It's causal relationship that we want.

		Okay, so let me talk about the scope of the Preliminary Quantitative
Risk Assessment.  By the scope, I mean, what questions does it set out
to address?  As I read the preliminary QRA, it seems to me that it
presents evidence of statistically significant exposure response
associations.

		I think we can and should quibble over how good is that evidence, how
strong are the associations, do we believe that they're really there? 
But the basic setup is to say, let's document all the reasons that we
think that in numerous studies there's an association, a positive
association and a nonrandom association between exposure and response.  

		That's the basic scope of the study.  And after going through hundreds
of pages of examination, the study concludes that reducing exposure
further would most likely reduce risk further.  And there's an
implication for policymakers that this is a good thing to do.  We want
to reduce risk.

		I want to point out to you that there is no logical bridge from "here
is an association" to "therefore, if you reduce exposure, health risks
will go down."  This is deriving a causal conclusion from non-causal,
i.e., association-based data, and that, there's no legitimate basis for
doing.  You need to use causal analysis if you're going to draw valid
causal conclusions.

		Key questions for evaluating -- excuse me, evaluating any
quantitative risk assessment are, first, are the presented results
valid?  Do we believe them?  And what I mean by valid, in this case, is
do the results flow from the data?  Do they follow from the data, or are
they interpretations imposed on top of the data, where somebody could
equally well have reached a different conclusion from the same data?

		Secondly, do the premises imply the stated conclusions?  And I've
already told you my conviction that from non-causal analyses, valid
causal conclusions do not follow.

		Thirdly, how shall we determine the answer?  And I'll just take the
shorthand of saying, we ought to apply appropriate methods.  They're
there; let's use them.  

		And, fourthly, how confidant can we be in the answer?  So I'll look at
those things.

		But let me begin -- I think everybody here -- most likely -- I'm
not going to ask for a show of hands.  But I would expect everyone to
agree that association is not necessarily causation, although if
causation is there, you might expect to see association.

		Let me remind you with a simple example, one of hundreds, of why this
distinction matters.  Here I've plotted on the x-axis the cell phones
per capita in different countries, and on the y-axis, the cardiovascular
mortality rate per capita in those same countries.

		And you can see that there's a roughly linear, non-threshold, clearly
statistically significant association between cell phones per capita and
heart disease per capita -- I'm sorry, cardiovascular mortality per
capita.  And yet I suspect that few of us would be convinced by this
association that reducing cell phones in a country would reduce
cardiovascular disease, right?

		So here we have association without causation.  We might have
excellent goodness of fit, and we do.  We might have a high R squared,
and we do.  What we're missing is, what would happen on the y-axis, the
health outcome, if we were to change the x-axis, if we were to force a
change by reducing exposure to cell phones?  That's the question that we
want to know the answer to.

		Common sense says, well, nothing would happen.  What I want to
emphasize is that there is nothing in the statistics and nothing in the
data that provides warrant for drawing a causal conclusion from
non-causal data.  

		And yet what OSHA has done, in effect -- and in fairness, not just
OSHA but many papers on which it relies, is exactly analogous to saying,
well, look, look at this nice, strong relationship.  Therefore, if we
reduce exposure, health effects, health risks will go down.  It's an
unwarranted conclusion.

		And this is true in key studies.  This is true of Attfield and
Costello, 2004.  It's true for lung cancer studies and for non-lung
cancer studies.  There's a real problem here, even in IARC's analysis,
which is the question, "How much of this association is causal?" is
never explicitly addressed or answered using reproducible, generally
acceptable methods.

		There is sort of an implicit assumption that the answer is 100
percent, the whole thing is causal, but it might be that 0 percent of it
is causal.  It simply hasn't been addressed.

		So, in terms of scoping, I would say, the scope didn't include -- the
scope of the preliminary QRA, excuse me, didn't include the question of
greatest interest to policymakers, one that should be of greatest
interest to policymakers, which is what would happen to health if we
were to change exposure standards?

		Now, there are many non-causal sources of association.  I want to
touch on a couple of them.  I notice that Dr. Park is in the room.  I
was struck, in reading the Park et al. paper from the early 2000s, 2002,
I think, that age categories of about 5 years were used, bins, to put
people into groups.

		It occurs to me that there's a phenomenon called residual confounding,
which is, if within a five-year bin -- older people tend to have higher
risks of lung cancer, for example, or any other disease, and if within
that same five-year bin the older people also tend to have had greater
exposure, then there will be confounding within the bin that is not
controlled for by creating different bins.

		So in other words, the words say we controlled for age by using
five-year age buckets, but in fact there can be residual confounding
that needs to be tested for and controlled for.  That would be one
possible source of an association that is not a causal association. 
It's just that people who are older both have more exposure and have
higher risk.  But that doesn't mean that one causes the other.

		Coincident trends -- over time, risks have come down and exposures
have come down.  This is another possible source of association without
necessarily causation.  Modeling assumptions are a very popular and
often hard-to-spot source of association.  I'll touch on that in a
moment.

		Model interpretation -- I show you here a picture from Steenland et
al., from 2001, which the authors suggest is "reasonably
monotonic" -- their phrase -- and which they interpret as not
providing evidence for threshold.  

		And what I'll say is if you look carefully at the graph, you'll see
that, in fact, there is a negative slope at low exposures, and again at
higher exposures.  Whether that's reasonably monotonic, I don't know,
but it's not monotonic.  It goes down before it goes up, although on
this scale that's a little bit hard to see.  You have to look carefully.

		But I'll also point out to you that if you thought there was a
threshold, it would be hard to imagine a dataset that gave stronger
support.  I mean, you could see there's really no increase in risk until
exposure gets sufficiently large.

		So I think there's a lot of just interpretation that gets put on
graphs and data in the OSHA key papers that somebody else could have
interpreted differently, and perhaps no less validly.

		Let me turn to the meatiest slide that I have here today.  I don't
want to go through all of these in equal detail because some of them
have been touched on already.  But I do want to look at, not questions
of scope, but questions of method.

		So a question of scope is did we ask and answer the right question.  I
think the right question is a causal question, which has not been
addressed.  I think the question of method is, have we used correct
methods to at least answer the questions that we formulated?  And here I
think there are significant problems.

		So, the overall theme here is one that was touched on more than once
this morning, which is that selection of studies and data and models and
methods and assumptions can create uncontrolled false positive
associations.

		Now, again, it's fair to say -- well, couldn't they also create
uncontrolled false negative?  And in the abstract, perhaps the answer is
yes.  But in the concrete, I think there's a steady push towards false
positives, as I'll now explain.

		On the study selection side, there has been a fair amount of
discussion already about whether the studies that OSHA thinks are most
credible, are most credible.

		Now, I'm familiar with the concept of deferring to regulators on
topics that they should be deferred to on.  But I do think it's fair to
say that if OSHA, or any regulator, sets out a set of principles that
they're going to use for choosing among studies, then they should abide
by those principles and not pick and choose some over others.

		In my written testimony, I have elaborated what seems to me to be
cherry-picking by OSHA where, for example, they preferred and used and
gave great weight to the Attfield and Costello 2004 study, even though
that study didn't meet the criteria that they had specified in terms of,
for example, having good measurement -- sorry, exposure estimation
data.

		And they rejected other studies, prominently Pam Vacek's study, that
to some people, at least using the criteria that OSHA had specified,
might seem to have been at least as good and probably better.

		My role as a commenter is not to say which is the right study, but it
is to underline that if criteria are going to be developed and used,
then they should be applied consistently.  Failing to do that creates
the potential for study selection bias.  And I think that potential is
manifested in the preliminary QRA.

		Data selection biases, this is another vexed topic that a lot of
people have spent a lot of time talking about.  The question is -- or a
question, which I'm not going to answer, is, is it legitimate to throw
out data, let's say at the high exposure levels, not only because it
doesn't show what you expect it to show, for example, because risk comes
down at high exposures and that just doesn't seem to make sense, but is
it legitimate to throw out that data if you have good reason to suspect
that that high exposure data isn't very trustworthy?

		Well, without being able to answer that question, is it a good idea to
throw out data, what I can say is if you are willing to throw out data,
then the significance levels that are quoted for whatever results you
come up with must be adjusted to reflect the fact that you've taken more
flexibility in your approach to data handling than the tests assume.

		The tests of significance don't provide for selectively excluding data
in order to get a coherent story.  And if one is willing to do that,
then one ought to pay the price of reducing p-values accordingly.

		On the data selection front, Park et al. excluded some high exposure
data.  Attfield and Costello excluded some high exposure data.  And they
did so not in an ad hoc way, and not simply to get a good fit, but they
did so because they thought there was a lot of exposure measurement
error at the high exposure levels.  And that may well be true.

		But if it is true, not only does one have to adjust the p-value for
tests after throwing out some data, just to keep the statistical
accounting honest, but then one ought to use methods and models that
explicitly take account of, model exposure measurement error in the data
that remains.  

		It is a big, important factor and needs to be taken in account.  It's
not hard to do.  And it hasn't been done, either in the OSHA preliminary
QRA or in the key studies on which that QRA rests, including Attfield
and Costello, and Park and Rice and Steenland.  Not one of them used
correct methods, and these are not difficult methods.

		So that's on data selection.  On model selection and multiple
comparisons, I think we've all noticed that there's a lot of choice in
the modeling that's been done.  So different exposure metrics have been
tried, sometimes lag, sometimes not, sometimes cumulative, sometimes
something else.

		Different lag lengths have been tried, with inconsistent results
coming from different studies.  Different model forms, whether loglinear
or linear relative risk or something else have been tried.

		This multiple testing in order to see whether something looks
significant and coherent and easy to explain again carries with it a
necessary price if one is going to do valid statistical inference.  

		And the necessary price is to understand that the reported
significance levels, the reported p-values are no longer valid in the
presence of multiple testing and need to be adjusted.

		The model form -- I've shown here, on the right-hand side, I've drawn
a picture in which there is no relationship between x and y.  What I did
is I just randomly sampled x values, and independently I randomly
sampled y values, and I created a random scatter plot.  But then I fit a
model to that data.  My model was risk as proportional to exposure, plus
some error term.  

		And if you go home and do that, using your favorite statistics, what
you will find is that there is a statistically significant positive
relationship between these two independent variables.  How can that
possibly be?

		Easy.  If risk is proportional to exposure, that means when exposure
is zero, risk has to be zero.  So I've pinned down one end of the line
to go through (0,0).  And risk is never negative, and exposure is never
negative, and so all of the other values are positive.  

		Well, if you pin one end of the line down at (0,0) and all the other
data points are positive, you can be guaranteed that that line will
slope upward.  

		I show you this simple example because it illustrates in textbook form
the problem of model specification error, making up a model that might
seem intuitively plausible to you, but it has not been validated, and in
fact it leads to biased results.

		OSHA, as far as I know, has not used this particular model, but
neither -- and when I say OSHA, again, I mean here, the papers on which
OSHA relies, they've not presented responsible model diagnostics in
keeping with sort of good housekeeping practices for epidemiology and
statistics.  They haven't shown residual plots.  They haven't said, yes,
this model describes the data generating process pretty well.

		They have used goodness-of-fit comparisons, which are almost
meaningless.  Goodness of fit is a very weak criterion.  But there
hasn't been any thorough justification for models. 

		And sometimes, going back to the original papers, it's been impossible
for me to figure out exactly how the models were chosen at all, even
after carefully reading the underlying papers, because they contain such
elusive words as "adjustments were made and a final model was selected
based on visual evaluation."  I can't reproduce that.  I don't
understand what it means.

		The result of picking any model, including the best-fitting one, and
then assuming that it's correct is to underestimate confidence
intervals.  They're too narrow because model uncertainty has been
ignored.

		This issue of ignoring model uncertainty is a problem for Rice et al.,
2001; for Steenland et al., 2001; Attfield and Costello.  None of the
key papers that I looked at, nor QRAs just -- sorry, nor OSHA's
preliminary QRA, in its discussion, provided a thorough treatment or any
treatment of model uncertainty and its effects on what the confidence
intervals for its risk estimates should be.  It's a grave omission.

		Okay.  Proceeding now more quickly, model over-fitting refers to using
the same model -- or I'm sorry, the same data, both to fit a model and
then to assess the goodness of fit.  It's well known that that leads to
an overly optimistic assessment of how good the model is.

		Biases due to residual confounding, we've already touched on, so I
won't say more about them.  Model uncertainty bias have -- I'll -- I
won't say more about it now.  Biases from exposure estimation errors,
because they were discussed this morning, even though I think they're
very important, I'll save for a final discussion of thresholds.

		Because to me, the question of is there a threshold there interacts
strongly with the question of have we modeled uncertainties in exposure.
 So I'll come back to that.

		JUDGE SOLOMON:  So you're at uncertainty characterization?  So could
you do this in about five minutes, and then we could have some questions
from the peers and then the public and then OSHA?

		DR. COX:  Yes.  I think that'll fit very well.

		MR. KING:  He's not at uncertainty characterization quite yet.

		DR. COX:  Yes.  Okay.  So the next question is are these complaints
nitpicking?  Is this -- no study is perfect.  Is this merely
highlighting imperfections, or are they essential?  And I think the
answer is that they are essential.  And in the interest of time, I won't
double-click on each of these things, but I'll give you my bottom line.

		If you ask how will risk change when exposure is changed, and what's
the ratio of change in risk to change in exposure, I believe that the
bottom line is that the answer is unknown.  There's nothing in the
preliminary QRA that provides a credible answer to that question, for
reasons similar to my cell phone example.

		Let me move on to the question of thresholds.  OSHA's analysis and the
supporting studies, I think, do not show that -- well, it's hard to
show a negative, but they don't show the absence of a threshold even at
relevant exposure levels, for example 100 µg/m3 or above, for the
cancer and non-cancer endpoints of interest, nor do they make absence of
a threshold plausible.

		A big reason for this is that they don't adequately model errors and
uncertainties in exposure estimates.  And those errors and uncertainties
can hide a threshold if it's there.

		I will say that there are, in my judgment, good reasons to expect a
threshold on biological grounds, and epidemiological data are not
inconsistent with there being a threshold.  Again, I'm showing the
picture of Steenland et al., 1991, which is not inconsistent with there
being a threshold.

		But OSHA did look quite hard at this question for lung cancer.  They
looked at using an incorrect or an inappropriate technical tool of Monte
Carlo estimate -- sorry, Monte Carlo simulation.  Monte Carlo
simulation does not address the problem that a correctly specified model
must include an error term for exposure.

		What it does is to replace that problem, fitting a curve when exposure
is uncertain, with another problem, fitting a curve when exposure is
soon to be known, and then it repeats that many, many times.  It draws
many know exposures from a distribution of exposures.

		That's not the same thing as correctly modeling the effects of
exposure measurement error.  So although they spent quite a lot of
effort there -- I mentioned Berkson error, for example -- but I don't
think that they correctly addressed the question.

		So now, and indeed I'm at uncertainty characterization.  I can be
swift.  Model uncertainty has not been quantified.  There's no base in
model averaging.  There are no other techniques here that correctly
characterize model uncertainty or its effects.

		Uncertainty due to errors in exposure estimates has not been
quantified.  An effort to do that for the special case of cancer
endpoints was made, but used again an inappropriate tool.  It's one that
doesn't solve the problem.

		Uncertainty about the causal fraction of associations has not been
quantified.  There's sort of an implicit assumption that it's 100
percent, but there's no explicit discussion of what the fraction is or
what the basis is for assessing the fraction.  And that, to me, is a
huge hole in the analysis.  You can't -- you basically can't answer the
question of concern to policymakers without filling that hole.

		And then uncertainty due to various biases including residual
confounding has not been quantified.  

		So, in conclusion, on uncertainty characterization, the preliminary
QRA results, to me, have no known relation to the real world.  We don't
know how exposure and response to causal relation in the real world
looks, based on what's been done in the preliminary QRA.  That's why
it's not ready for use, in my judgment.

		So let me then wrap up with the following recommendations.  Fix what's
broken.  Fill the holes that are missing where they really matter.  And
that's basically the five things that I've talked about.  

		So we ought to correct for biases.  We ought to do causal analysis,
very importantly.  We ought to model exposure errors.  We ought to
reexamine the question of thresholds taking exposure errors into
account.  And we ought to present an uncertainty analysis that includes
model uncertainty as well as exposure uncertainty, and explain whether,
when that's done, there's any evidence of an association remaining. 
Thank you.

		JUDGE SOLOMON:  Thank you.  Ms. Ryder, do you want to enter your
appearance?

		MS. RYDER:  Sure.  My name is Anne Ryder from the Solicitor's Office.

		JUDGE SOLOMON:  Okay.  Now, any of the peer reviewers, do you want to
ask some questions?  Okay, so let's do this in order.  Why don't you --
you haven't asked any questions yet.  Would you go forward and state
your name, please?

		DR. MILLER:  My name is Brian Miller.  I am from the Institute of
Occupational Medicine in Edinburgh.  I'm an epidemiologist and
statistician.

		Dr. Cox, you're very critical of the analyses that have been done.  I
look at the epidemiology of silica and silicosis over many decades, and
I think everything that we've done can be criticized on the basis that
you have criticized the current work.

		Are you unconvinced that silica causes silicosis?  And have we been
wasting our time in bringing the limits down and protecting people from
silica?

		DR. COX:  No, I'm not unconvinced that silica causes silicosis.  And I
don't know whether or for how long we might have been wasting time.  If,
indeed, we're already below a threshold, and have been for some time,
then recent reductions won't have purchased any more -- I'm sorry,
recent reductions of exposure won't have purchased any improvements in
health.

		JUDGE SOLOMON:  Okay.  So we have two other people raised your hands? 
Go ahead, doctor.  Dr. Ginsberg, if you want to get into --

		DR. CRUMP:  Kenny Crump's my name.

		JUDGE SOLOMON:  -- position, maybe that will save a few seconds.

		DR. CRUMP:  Good afternoon, Dr. Cox.  Good to see you again.

		DR. COX:  Good to see you.

		DR. CRUMP:  I have several questions.  I'm not sure I've prioritized
these appropriately.  You were very emphatic that OSHA should use causal
studies and causal analyses.  Can you tell OSHA of any causal studies of
silica that they should have used but did not use?

		DR. COX:  I think OSHA could look at a paper from around 2007 of
Brown's, on some of the issues and causal analysis, but I think that the
crystalline silica area has been behind other particulate matter areas,
such as PM2.5 epidemiology, in not using causal analysis methods.  So
no, I can't point to a good study that they should have included but
didn't.

		DR. CRUMP:  Thank you.  Well, you referred several times to the graph
of Steenland, and you said that this is not inconsistent with a
threshold.

		DR. COX:  Yes.

		DR. CRUMP:  Couldn't you say that about any study?

		DR. COX:  Perhaps I should have clarified and said it's not
inconsistent with a threshold at or above the current PEL.  And no, you
couldn't say that about just any study.

		DR. CRUMP:  You could not?

		DR. COX:  No.  I think that since it's impossible to prove a negative,
empirically, you can't say if you go low enough for any substance, might
not there be a threshold there?  You could never rule out that
possibility.  And I will grant you that.

		But here I'm making a stronger statement, which is not somewhere
between here and zero, but somewhere between here and infinity, we may
already be below the threshold.  This is not a small hypothetical
threshold that I'm complaining about; it's a threshold that we may
already be below in practice.

		And I think that there are plenty of chemicals for which the
hypothesis of a threshold exists at or above current standards could be
ruled out because you see people getting sick at current levels.

		DR. CRUMP:  Well, would you say the other side of that question is
also true, is any data are not inconsistent with a no threshold?

		DR. COX:  I think until exposure errors are well modeled, the data as
interpreted via the current set of models are consistent with many
hypotheses, including both that there is a threshold and that there
isn't.  I don't think that the data, as interpreted through models that
don't take into account exposure measurement error, tell us what the
answer is.

		DR. CRUMP:  Okay.  In that graph that we were talking about,
Steenland, you graphed the x-axis in a log scale.

		DR. COX:  Or I didn't, but he did, yes.

		DR. CRUMP:  Okay.  But that does tend to make most any graph look like
it has a threshold, does it not?

		DR. COX:  No.  If you take an exponential relation and graph it on a
log scale, you'll get a straight line.

		DR. CRUMP:  Okay.  Do I have time for another question?

		JUDGE SOLOMON:  Depends on Dr. Ginsberg.

		DR. GINSBERG:  No, go ahead.

		DR. CRUMP:  In your written testimony, you testified, mentioned
several times a review by Health Canada.

		DR. COX:  Yes.

		DR. CRUMP:  And they demonstrate that a threshold approach to risk
assessment is appropriate for silica-induced lung cancers.  And then
later you interpret that as findings of Health Canada previously
discussed showing that a threshold is likely inappropriate on biological
grounds.  Does the threshold approach necessarily imply a threshold?

		DR. COX:  I'm not sure.  Those are Health Canada's words, but I think,
in fact, Health Canada did assert the plausibility of a threshold, as
well as advocating a threshold approach.  So I would defer to them for
exactly what they mean, but I think they asserted both.

		DR. CRUMP:  It is true that EPA, for example, would often use this
so-called approach without necessarily implying that there is a
threshold.  Is that right?

		DR. COX:  Well, again, if the proposition is that EPA would often
apply Canada's threshold approach, I would defer to the people who are
doing that work.  I'm not really sure.

		DR. CRUMP:  And would you agree that this approach -- one more,
please sir, that this approach would not be appropriate for OSHA because
it's not risk-based?  And OSHA is -- must -- requires estimates of
risk.

		DR. COX:  No.  I don't think I would agree with that, because the
whole question about what is the risk turns on is there a threshold?

		DR. CRUMP:  Thank you.

		DR. COX:  Thank you.

		JUDGE SOLOMON:  Thank you.

		DR. GINSBERG:  Okay, my question is going to --

		JUDGE SOLOMON:  State your name.

		DR. GINSBERG:  I'm sorry.  Gary Ginsberg.  I'm on the peer review
panel.  Two directions for my questions: one is I'm a little dumbfounded
about the concern over causality, given all the animal evidence that --
and, again, I can understand your cell phone example, and your baby
aspirin examples, where there's an obvious explanation for the
correlation.

		And you may bring up radon exposure or asbestos co-exposures or
smoking co-exposures to explain some of the associations seen in some of
the epidemiology.  What you didn't talk about are the confounders that
may have gotten in the way.

		But we have animal studies and we have cell cultural studies that show
the effects of crystalline silica on biological systems, which help
explain -- and usually toxicologists use the animal evidence in a
causal way to support the epidemiology.

		So my first question is what do you make of the animal evidence, and
why is it you do have such a strong focus on causality as an issue when
we have such a strong -- as IARC pointed out, and as there's a lot of
animal data to support the cause and effect of silica on lung pathology?

		And then the second direction is, because I don't know that much about
measurement error and its implications on statistics, is it possible
that if the corrections were made for measurement error, that the
regression slopes would be strengthened, because my understanding about
errors tend to bias towards the null and weakened associations.

		DR. COX:  So let me address those in reverse order.  The proposition
that omitted or unmodeled measurement error biases result specifically
in regression coefficients towards the null is correct when there's only
one explanatory variable, or one independent variable.

		It is incorrect, as I think I heard before lunch, in multi-variant
models.  And here we're dealing with multi-variant models.  So the
correct answer, to me, is not to speculate about what would happen if we
were to correct -- make the needed corrections; it's let's make the
needed corrections and see what happens.  A priori, and at an abstract
level, it could go in either direction. 

		What is -- what can only go in one direction is that when ignoring
uncertainty or failing to model uncertainty artificially reduces
confidence intervals, there is a bias towards false positives.

		Now, for your question about the animal data, to me, the animal data
are very coherent with human data in suggesting that, indeed, high
concentrations of various poorly soluble particulates can lead to
chronic inflammation, unresolved inflammation of the lung, and to a high
ROS lung environment that is conducive to the onset of fibrosis and
other lung diseases.

		However, the animal data are also very coherent, let's say insistent,
on the fact that such processes are threshold or threshold-like
processes.  One particle will be cleared, where many particles might
lead to a lung overburden or a unresolved inflammation situation.

		The relevant causality that I am emphasizing so much, and that took
you aback so much, has to do with is there a causal relation between
reductions in the current PEL and reductions in occupational risk. 
That's the causal relationship of interest.

		I don't think that it's usefully informed by animal models, which show
why exposing the lung to high doses causes lung disease.  I think we
understand that.  It doesn't illuminate the question of what happens at
ambient levels at currently permitted levels, or at lower than currently
permitted levels.

		JUDGE SOLOMON:  Before you ask the next question, are there any
members of the public who want to ask a question?  Okay, I see one hand.
 Would you please get in line?  And it's Dr. Park again.  Go ahead,
Dr. Ginsberg.

		DR. GINSBERG:  Yes, so are you aware that there is an animal-based
model system which incorporates dose-response data on the critical
loading dose for silica in the lung, which projects to a human threshold
for, again, critical loading capacity in the lung before you start
getting inflammation and lack of clearance that is well below the PEL?

		And, in fact, if you factor variability in, that study's projection is
that it's about 10 times below the PEL where some workers would have an
impact that would lead to inflammation.

		DR. COX:  Are you referring the Kempel et al. study?

		DR. GINSBERG:  Exactly.

		DR. COX:  Yes, I'm quite familiar with it.  And here the single most
important word in your question is the word "project," how that
projection is done.  And the Kempel et al. study did not account for
exposure measurement error in estimating human parameters from which to
extrapolate from animal data.

		So, again, I think we're right together, that high dose data in animal
and humans is firmly in support of inducing diseases at high doses.  But
I don't think that that's informative about what will happen to health
when we reduce exposures below current levels.

		DR. GINSBERG:  Yes, I brought up that study because it has both low
dose and high dose.  The short time frames were low-level loading, where
they didn't even -- did not see a no effect level at the lowest
loadings.  So I think it's worth taking a harder look at that --

		DR. COX:  Again, low compared to other doses used in that study, but
not low compared to what many people are exposed to.

		JUDGE SOLOMON:  State your name.

		DR. PARK:  Robert Park from NIOSH in Cincinnati.  It's ludicrous to
hear someone question causality.  There's 100 years of research in
occupational medicine, in exposure assessment.  People here even in
industry would agree that silica they say causes silicosis, which causes
lung cancer.

		There's some debate about whether the middle step is required. 
There's no question that there's excess lung cancer in silica-exposed
populations.  We look at literature, and we identify what we call good
studies.  Good studies are ones that look at confounding, asbestos,
whatever.  We make judgments.  If there's data that allows one to
control for confounding, that's part of the analysis.

		If there is confounding that we can't control for, we evaluate it.  We
ask how bad could it be?  There's a lot of empirical judgment from
people who know these populations, know these exposures, know these
industries, who can make very good judgments about that.  We aren't
stupid.  So I don't know where you're coming from.

		JUDGE SOLOMON:  Okay, let me -- two things.

		DR. PARK:  Number two.

		JUDGE SOLOMON:  No, wait a minute.  Stop at number one.  First of all,
you were nodding with -- just for the record, the witness was nodding
along when Dr. Park was making some of the statements.

		And, number two, is that a question mark at the end of number one?

		DR. PARK:  Yes.

		JUDGE SOLOMON:  Okay.  You can answer the question, Dr. Cox.

		DR. COX:  Okay.  I was nodding as I ticked off the various points:
we're not stupid, people have looked at this for a long time, we make
judgments, and so forth.  However, it has been abundantly shown -- are
you familiar with the Wason selection task?

		DR. PARK:  No.

		DR. COX:  No.  It has been abundantly shown by psychologists who have
studied the capacity of expert scientists and statisticians to make
judgments about causality, that we typically have very high confidence
and very low -- or not as high as we would like -- accuracy in our
judgments.  Judgment is not a substitute for rigorous analysis.

		I do not question that at sufficiently high exposures, there are real
effects.  I do question that those effects can be extrapolated without
careful causal analysis, to the specific context that's of interest to
us today, which is will further reductions in exposure bring about
further health benefits to workers?

		I think that answering that question soundly requires a great deal
more than judgment.

		JUDGE SOLOMON:  Okay.  How long is this going to be, Dr. Park, the
next question?

		DR. PARK:  Two more, pretty brief.

		JUDGE SOLOMON:  Okay.  Here's the problem.  We have questions that are
going to come from OSHA, and then we have another two-hour presentation.
 So we're supposed to end at 5:00.  Apparently -- we could go on.  I
mean, if these are that crucial, we could go on.  So if you have two
more questions, how long is it going to take?

		DR. PARK:  Two minutes.

		JUDGE SOLOMON:  Okay, go ahead.

		DR. PARK:  Okay, on the question of model selection, again, it's not a
simple-minded "Let's pick the best one."  There's judgment involved. 
There's biology.  There's prior observation in other studies.  Linear
exposure responses actually have a lot of prior support.  We don't just
pick the one with the best estimate.  You might want to respond to that.

		Finally, on the question of exposure error measurement and confidence
intervals, that's true, but we're actually not interested in confidence
intervals.  We've already established causality.  We need to estimate
what is our best guess for the exposure response.  What's our best
estimate?

		It doesn't matter if the confidence interval is super tiny or somewhat
wider.  We don't have any choice.

		DR. COX:	So, I would disagree with you that causality has been
established at relevant exposure levels, and I would disagree with you
with confidence intervals don't matter.

		DR. PARK:  Okay.

		JUDGE SOLOMON:  Okay.  So I assume OSHA has questions.

		MR. PERRY:  Yes, we do, Your Honor.  If I may, Tiffany DeFoe has a
question for Dr. Cox.

		MS. DeFOE:  Thank you.  

		Dr. Cox, I'd like to get a better grasp on your thoughts about the
sort of uncertainty in error analysis that you describe, and what they
might mean for OSHA decision making.

		So earlier in your statement you made a general statement that until
exposure errors are well modeled, many datasets might be consistent with
a threshold effect or with no threshold effect.  

		So specifying that a little bit to say that if you believe that until
the exposure errors are well modeled in this case, the current data are
consistent with a threshold or no threshold in the area of interest, is
that something you'd agree with?

		DR. COX:  If I understand you correctly, yes.  And just to be clear,
I'm referring here to models based on epidemiological data.

		MS. DeFOE:  Yes, thank you.  And when you discuss the procedures that
you say we should do to account for types of error in exposure, are you
suggesting simulation studies that we -- okay.

		DR. COX:  No.  Here is an incorrect procedure.  Use simulation to draw
a number of imputed exposures from some uncertainty distribution.  Then
fit a model that does not account for exposure measurement error to the
drawn values.  Repeat it many times and see what distribution of
regression coefficients you get.  That would not be a correct procedure.

		Here is the heart of the matter.  The basic approach to modeling, when
exposure error is ignored, is to say, we can regress risk against
exposure and come up with some coefficient.  How big is that
coefficient?  Right, that's the basic idea.

		The basic idea that would be correct is to regress risk against
estimated exposure plus error.  So in other words, you put an error term
right into the exposure of your model.  That calls for a different
fitting procedure and will produce different results.

		So what I'm saying is we have to do that second thing, because doing
the first thing a number of times, for example, in Monte Carlo
simulation, doesn't solve the basic problem, that you haven't correctly
specified the model to take exposure estimation error into account.

		MS. DeFOE:  Okay, let me make sure I understand.  So you're suggesting
that we should model error --

		DR. COX:  Correct.

		MS. DeFOE: -- of various types and should, based on different
assumptions and information about the nature of the error, fit a series
of different models to see the best model form?

		DR. COX:  That's going a little bit further than I did.  I would --
all that I'm insisting on, if you will -- actually, I'm not in a
position to insist, but all I'm strongly advocating is that the models
that are fit should include errors in the independent variables.

		So repetition of a model that doesn't have errors in exposure or other
independent variables won't solve the problem of correctly estimating
slope coefficients for a model that does have error in the exposure
variable.

		So what I'm saying is use those models that include error in the
exposure variable.  And then whether you do a series of them or one, I
haven't gone that far.

		MS. DeFOE:  I'm going to pass the microphone on.  I might come back if
I manage to reformulate my question correctly.

		DR. COX:  Great, thank you.

		JUDGE SOLOMON:  So who is next, Mr. Perry?

		MR. PERRY:  Next will be Janet Carter from our Office of Chemical
Hazards - Metals.

		MS. CARTER:  Good afternoon, Dr. Cox.  I appreciate the opportunity
to talk to you today and ask questions.  I wanted to get a little more
at your view on causality.  I think it's very interesting.  You
certainly state in your testimony and today that causation -- or
association is not causation, and I think that that's understood.

		You talk about some animal data being coherent with human data and
that it tends to lead the conclusion of a threshold response.  How do
you relate that to some of the in vitro studies which show direct acting
mechanism of genotoxicity to epithelial cells?  How would you evaluate
that in this causality model?

		DR. COX:  Yes.  I think I would be very slow to try to extrapolate
from in vitro and cellular level responses.  The fact that you can see a
NF-κB pathway activated at the cellular level essentially indicates
that this is a living organism that's responding to stimuli.  I mean,
these are highly conserved pathways, and they'll light up in response to
any number of shocks.

		Extrapolating that -- from that to a whole organism, I think, is not
prudent for lung cancer or other lung diseases, because I think that the
lung response is primarily a tissue-level response.  And what I mean by
that is that you have cross-talk between different cell populations.

		You do have your alveolar epithelial cells, but you also have T-cells
that have infiltrated the lung, for example, that play an important
mediating role.  You have interleukin-8 signaling to a variety of cells
within the lung.

		To understand the onset and the dynamics and the possible threshold in
the lung's response, I think it's fruitful to take a tissue-level view
in which multiple cell populations help to regulate each other in
feedback loops.  And that's not going to come out of a cell-level
perspective.

		MS. CARTER:  But there have been cellular studies looking at tissue
responses, and actually taking tissue constructs and doing the same
thing, and seeing that evaluation.  So that would seem to fit within
your paradigm, would it not?

		DR. COX:  I guess we'd have to talk about specifics.  I think that the
paradigm in which chronic unresolved inflammation of the lung is a first
step towards increased risk of various lung diseases has a lot of
support for it.  So inflammation plays an essential role.

		And as long as we have enough of the lung modeled to sensibly discuss
inflammation and its sequelae, I think we're in good shape.

		MS. CARTER:  It does, but there are expert authoritative bodies who
have not discounted a direct genotoxic mechanism.

		DR. COX:  Sure.  And again, as --

		MS. CARTER:  And certainly -- I mean, I don't think anybody would
argue the case for inflammation, absolutely not.

		DR. COX:  Right.  And again, I think it's hard to prove a negative,
especially at relatively low doses, but I think the best supported model
is the one we've been talking about, where in fact, without inflammation
you don't see excess risk.  I think there's good science that needs to
be done to clarify that.

		JUDGE SOLOMON:  You have any other questions?  Mr. Perry?

		MS. CARTER:  I think I'll leave it there.  No.

		MR. PERRY:  Yes, Dr. Cox, I just have a couple.  On Page 15, I think
you talked about this idea of residual confounding, but I don't see --
I don't really see a citation to that.  Is this something that's been
raised or discussed in the literature as a --

		DR. COX:  I think residual confounding has been raised and discussed
in the statistics literature, but I don't know that it's -- in fact, I
have an impression that it has not, just reflecting on the papers that
I've read, been raised specifically in the crystalline silica
literature.

		MR. PERRY:  Oh, I -- yes, I wasn't asking if it was specific in the
silica literature, but --

		DR. COX:  But in the general --

		MR. PERRY: -- in the context of epidemiology --

		DR. COX:  Yes.  It's well known.

		MR. PERRY:  Literature in general.

		DR. COX:  Yes.

		MR. PERRY:  If you could --

		DR. COX:  Provide reference?

		MR. PERRY:  -- provide some information to the record, I would invite
you to submit whatever papers you think is appropriate to illustrate
that phenomenon.

		Also, just so I'm clear, you also, on that same -- starting on that
same page, you say, "Quantitative risk assessment such as what OSHA has
done, they routinely use a large excess of false positive errors."

		DR. COX:  Right.

		MR. PERRY:  So, and then you go on to -- I'm not familiar with the
papers you're citing here, but you quote one of them.  I think it's --
looks like Ioannidis, 2005?

		DR. COX:  Oh yes.

		MR. PERRY:  As saying, "Most published research findings are wrong."

		DR. COX:  That's actually the title of his paper, yes.

		MR. PERRY:  That's the title of the paper, okay.  Wow.  I mean, the
implications are rather staggering to me, but my question here is
how -- first of all, I'd like to know really how common this is.  If
you've got -- again, I would invite you to submit any additional
studies of anyone who's actually analyzed scientific literature and
identified false positive studies and --

		DR. COX:  And indeed they have.  And actually I've cited those.

		MR. PERRY:  Okay.  So they -- you -- those are the ones you --

		DR. COX:  But there's also a -- there's a January editorial in
Science by Marcia McNutt, the new editor of science, that touches on the
related topic of non-reproducibility of studies.

		MR. PERRY:  Okay.  I'm wondering, as studies are done over time, as we
have in the case of crystalline silica, more and more studies look at
different working populations, studies find associations in different
situations.  More and more of them accumulate.  Doesn't -- wouldn't
that reduce the odds that really this can all be explained through a
false positive bias?

		DR. COX:  I wish that were true, but I don't think it is.  I think
doing the same basic thing a thousand times will tend to produce the
same result a thousand times, but that's not evidence that the result is
correct unless the methods and analysis are correct.  It shows
consistency, but not correctness.

		MR. PERRY:  So as I understand it, you basically think there's a good
possibility that the entire literature base, with respect to lung cancer
now, I'm talking about, is wrong?

		DR. COX:  You mean with respect to lung cancer in the context of
crystalline silica?

		MR. PERRY:  Yes, sir.

		DR. COX:  I think that consistent with the findings of Lauer and
Ioannidis and others, I think that it's very possible and plausible that
there is a consistent pattern of false positives in the literature base,
yes.  And that implies, yes, they are wrong.  False positives are false.

		MR. PERRY:  Okay.  Thank you.

		DR. COX:  Thank you.

		JUDGE SOLOMON:  Okay, we're right at 2:02 right now, so --

		DR. COX:  Okay.

		MS. RYDER:  Yes, I think we still have a few more questions, if that
would be okay.

		JUDGE SOLOMON:  Well, here's what we could do.  We have -- we're
probably going to have to take a few minutes to, you know, change the
cast of characters up here.  They've got approximately two hours, but
that includes the questions from the peers and from the public, and you
get to ask questions again of their witnesses.  They come back on Friday
this week.

		MS. RYDER:  The peer reviewers won't be here.

		JUDGE SOLOMON:  So -- the peer reviewers will not be here, right.  So
we probably better make sure that we cover, at a minimum --

		MR. PERRY:  Yes.  I mean, this is a very important witness for us to
ask.  We only have a few more questions left.

		JUDGE SOLOMON:  Okay, go ahead.

		MR. PERRY:  I think Ms. Ryder has a few.

		MS. RYDER:  Sure.  B.J., you can go first if you want.

		MR. PERRY:  Okay.  Then this is Brian Albrecht from Salt Lake City
Laboratory.

		MR. ALBRECHT:  Thank you.  

		Dr. Cox, in your written submissions to the docket, you wrote a paper
called "Statistical Assessment of Performance Tests for the Analysis of
Respirable Crystalline Silica Quartz by Commercial Laboratories Using
XRD."  I was a little surprised that you did not discuss any of this
paper or this data today.  Would you mind just summarizing very quickly
what the findings were and how they affected your testimony?

		MR. KING:  Excuse me.  That will be discussed by witnesses on our
panel next Wednesday, in depth.

		MR. ALBRECHT:  Oh.  Well, I'll save my questions till then.

		DR. COX:  Thank you.

		MR. PERRY:  Fine.  Well, Dr. Cox, will you be on that panel?

		MR. KING:  No.  He wasn't --

		JUDGE SOLOMON:  In that case, answer the question.

		MR. KING:  All right, go ahead.

		DR. COX:  I don't think it really played any role.

		MR. ALBRECHT:  So it had no bearing on your testimony today?

		DR. COX:  I think that's correct.  My testimony is based first on my
reading of the OSHA preliminary QRA, primarily on that, and secondly on
reading key papers that support that.  And I think that's really the
basis of my testimony.

		MR. ALBRECHT:  Okay, thank you.

		DR. COX:  Thank you.

		JUDGE SOLOMON:  Ms. Ryder.

		MR. PERRY:  Ms. Ryder.

		MS. RYDER:  Okay, I just have one question, Dr. Cox.  You talked a
little bit earlier about the false positives that are --

		DR. COX:  Yes.

		MS. RYDER: -- present with a lot of the studies on lung cancer. 
And, but I believe, in your comment you didn't say that there are any
of those same false positives with studies dealing with silicosis and
silica exposure.  Is that correct?

		DR. COX:  I don't think I opined on that.  So -- and I really haven't
looked carefully at the question.  I do take it as given that silica at
sufficiently high and prolonged exposures causes silicosis.  I've not
really examined that literature.

		MS. RYDER:  So you don't think that those studies have the same issues
that some of the lung cancer studies have?

		DR. COX:  I don't really know.

		(Off microphone conversation.)

		DR. COX:  Yes.  Neil is pointing out that there's a different question
about whether those issues arise in the context of discovering
thresholds and performing risk assessment.  However, the easiest answer
is that I haven't looked at the question of the connection between
silica and silicosis, nor do I think that the -- well, here's what I
really think.

		MS. RYDER:  That would be great.

		DR. COX:  You can't tell the difference between the results in the
OSHA preliminary QRA and the results that should be expected if there
isn't really a relationship there, but you apply these methods and
modeling to just purely random data.

		So if you walk up to the preliminary QRA and say -- and ask, does it
do a good job of showing that these conclusions are likely to be
correct, I would say no.  It sort of assumes that these conclusions are
correct, but it doesn't show it.

		Is that true for silica and silicosis?  I expect that the evidence is
much stronger for silica and silicosis.  But I haven't reviewed it, so I
can't testify to it.

		MS. RYDER:  Okay.  Are you planning on doing that before the -- in
the posting and comment period?

		DR. COX:  No.  I -- no.  I was really -- my comments are based on
the preliminary QRA and just trying to evaluate that.  And unless asked,
I'm not planning to do anything further.  Thank you.

		MS. RYDER:  All right.

		JUDGE SOLOMON:  Is that it?

		MS. RYDER:  I --

		MR. PERRY:  I believe that's it, Your Honor.  

		JUDGE SOLOMON:  Okay, so it's --

		MR. PERRY:  Thank you, Dr. Cox.

		JUDGE SOLOMON:  It's 2:07.  Thank you very much.

		MS. RYDER:  Before we go off the record, could I get a copy of your
written remarks to add to the docket?

		DR. COX:  They should be in the docket already.

		MR. KING:  His written testimony is -- full written testimony is
already in the docket.

		MS. RYDER:  Can I have his slides for them?

		MR. KING:  We could leave a copy of these slides for you.

		MS. RYDER:  Okay.

		MR. KING:  Yes?

		MS. RYDER:  Right, sure.

		MR. PERRY:  Great.

		JUDGE SOLOMON:  Do you have a copy?

		MS. RYDER:  Yes, but I already put it in --

		JUDGE SOLOMON:  So do you want to mark it?

		MS. RYDER:  Yes.  I think this is going to be marked as Exhibit 7.

		MR. KING:  Exhibit 7?

		JUDGE SOLOMON:  And without objection, I will enter it into evidence. 


(Whereupon, the document referred to as Hearing Exhibit 7 was marked and
received in evidence.)

		JUDGE SOLOMON:  Okay, we're going to go off the record now.  We'll
come back in five minutes.  In the meantime, we will have a change of
the cast of characters here, and we'll have a brief discussion about how
we're going to handle the remainder of the day.

		(Off the record.)

		(On the record.)

 		JUDGE SOLOMON:  We have the Chamber of Commerce, Panel Number 2 this
afternoon.  As I said earlier, they had been allocated two hours, but
included in that allocation would be questions from the peer reviewers,
questions from the public, and then questions from OSHA.  So we're
probably going to have a presentation, hopefully of about an hour to an
hour and 15.  Is that okay?

		MR. CHAJET:  We'll give it our best shot, Your Honor.

		JUDGE SOLOMON:  Okay.  All right, so do you want to introduce your
panel, please?

		MR. CHAJET:  I do.  Thank you very much.  Good afternoon, everyone. 
Testifying for the Chamber this afternoon is Dr. Thomas Hall. 
Dr. Hall has more than 30 years of experience in industrial hygiene and
occupational health and epidemiology, primarily dealing with exposure
assessment.  He served on a faculty at the University of Oklahoma.  

		He's published research on exposures to airborne particulates such as
asbestos, diesel exhaust, hazardous air pollutants, crystalline silica,
and respirable coalmine dust.  Additionally, Dr. Hall has designed and
implemented large field studies to assess asbestos exposure variability
in silica exposures, mercury exposures, and lead exposures.

		We also have on our panel, testifying today, Robert Lieckfield. 
Mr. Lieckfield is a Certified Industrial Hygienist.  He is perhaps one
of the most well known CIHs in the country.  He directs Bureau Veritas
HSE Lab in Novi, Michigan and Zurich, Illinois, and Atlanta.

		He does industrial hygiene consulting around the world, project
management involving multi-disciplinary teams, and he consults for the
industry, the government, and academia.  Most importantly,
Mr. Lieckfield is very familiar with measurement and variability in
error and has audited analytical labs around the nation and led the AIHA
Lab Accreditation Program.

		In addition, we have testifying today Dr. William Bunn, who's sitting
behind me.  Dr. Bunn is a recently retired Vice President of health,
Safety and Security and Productivity and now conserves as a consultant
to Navistar.

		Along with his extensive experience in the corporate sector, Dr. Bunn
is a faculty member at Northwestern Illinois, University of Cincinnati,
and former faculty member at Duke, Yale, and the University of Colorado.

		And finally today, Dr. Gerhard Knutson will be providing testimony
based on his 35 years of experience as a consultant, evaluating and
designing industrial ventilation and air pollution systems and
conducting industrial hygiene studies for private industry, commercial
facilities, and for government clients.

		Dr. Gerhard is well known to all of you that have been involved in
EPA or prior OSHA ventilation issues and has testified on both the
government and design engineering controls around the world.  With that
I turn it over to our first witness.

		DR. HALL:  Thank you, Henry.  They picked me to go first because I've
been in the longest and forget the most.

		JUDGE SOLOMON:  Okay.  Would you state your name again, please?

		DR. HALL:  Yes.  My name is Thomas Hall.  You need my location?  I
live in Oklahoma, as a result of being a former professor of the
University of Oklahoma.

		I'm going to talk about a series of issues that deal specifically
with -- the first one is the change in convention from the ACGIH to the
ACGIH/ISO/CEN particle penetration conventions for determining or for
defining respirable dust. 

		I have a few brief comments on that, then I will go on to discuss what
the Occupational Safety and Health Administration has talked about with
some issues associated with the sampler performance and sampling for
crystalline silica aerosols, specifically focusing on the cyclones that
they have recommended be used, so they can -- to collect, excuse me, a
larger fraction of dust than can be collected typically with a sampler,
and separated like a cyclone 1.7 -- excuse me, a Dorr-Oliver cyclone,
operating at 1.7, because there are issues with that particular system
when you get down to aerosol concentration ranges of crystalline silica
at 15 µg/m3 and below.

		Relatively recently, at least in my history, there has been -- there
was a move, has been a move, and the OSHA proposal for quartz has
incorporated this move, is the adoption of a new particle penetration
convention.

		For the first 25, 20-some-odd years of my professional life, the ACGIH
particle penetration convention for defining respirable dust was kind of
the ruling convention.

		And more recently there has been a move to harmonize this
understanding, at least our understanding in the United States, of what
is respirable dust.  

		And with those in other places like in Europe, the British had a
different standard.  The French and Europeans had a somewhat different
standard.  Those have all been kind of combined now into one specific
convention, which when in comparison to the previous ACGIH convention,
in the review by Soderholm, a NIA scientist, demonstrated that there are
some significant advantages to this association, or this harmonization,
excuse me, and there are some built-in -- there is one built-in
difference that can make a significant difference when interpreting
exposure monitoring results collected under this new convention.

		And that is that there's approximately a 30 percent positive bias
built into the movement from the previous convention into this one.  And
this positive bias is built around the issue of what kind of particles
it collects.

		The adoption of the ACGIH/ISO/CEN particle penetration convention will
result in this 30 percent increase, as I just stated, in mass collected
by the proposed samplers, the OSHA proposed samplers.  This would
produce a difference in exposure results from the current exposure
results from historical measurements that have been used in the risk
assessments.

		That means the exposure measurements that are collected now really
don't represent necessarily the same exposures or risk that have been
characterized historically.

		The adoption of the new definition of respirable dust will not be
founded on the respirable fraction of dust that is thought to result in
disease risk, but will permit collection of dust more heavily weighted
to larger particles that are not thought to be part of the regulated
risk historically.

		The proposed imprecise definition, what is respirable dust, combined
with the device over-sampling of large, non-respirable particles will
result in an overestimation of respirable dust concentrations under
measurements taken under this convention.

		To move on with -- under this new convention, and under the issue of
having to collect enough quartz for analytical purposes, OSHA
recommended that a series of samplers that were called high volume
cyclone samplers be utilized or can be utilized, such that the sampler
can collect enough dust to feed into the analytical system that they can
give them a reliable analytical result.

		At low levels of quartz exposure surrounding the action level, the use
of a -- or the Dorr-Oliver in a 1.7 flow rate per minute -- 1.7,
excuse me, cubic centimeters per minute -- 1.7 liters, excuse me, per
minute, flow rate results in relatively small concentrations of quartz
on the filter for analysis.

		And it puts you into a range where the analysis is fraught with a lot
of uncertainty.  The use of these new samplers -- and there's a series
of them, but the one that was prominent in the discussion in the OSHA
PEA was a sampler called the GK 2.69. 

		One of the reasons it was kind of a prominent in the OSHA PEA is that
it operates -- when they were looking at their analysis, at between 4.1
and 4.2 L/min.  

		This would suggest that using this sampler at a bare minimum, if the
results are similar to what could be obtained in a -- or from a cyclone
operated at 1.7, with obvious scale-ups for volume, that this device
would just provide more dust for analysis.

		But recognized experts in the field of industrial hygiene have
demonstrated that there are potentially large contributions to the
errors associated with the sampling analytical process that could be
attributed to specifically the use of a high-volume sampler.

		Now, while there are, you know, a series of errors that basically are
well known, that is the errors associated with the analytical process,
and the errors that are potentially associated with the pump,
recently -- well recently, in the last 15 years have become relatively
minor given that most pumps are controlled by a computer microprocessor,
and the flow rate varies not significantly really.

		The potential errors, though, associated with the sampler, that is
this, specifically this GK 2.69 cyclone, have been demonstrated by Lee
and Stacey and Chen and others to be fairly significant and be dependent
upon the aerosol that you're sampling.

		In samples that have a relatively low GM, that is geometric mean --
or mean diameter, excuse me, and a relatively small GSD, the GK 2.69
sampling at about 4.2 L/min had some bias.  The bias was between 10 and
20 percent, a positive bias.

		But as the change in aerosol concentration that you're sampling, the
change in aerosol characteristics, as that changes, then that bias can
increase rather dramatically.  In aerosols greater than -- having a
geometric mean of 10 microns or greater, that bias can jump up to 20 to
60 percent.

		And that bias impacts the total variability that's seen in the
sampling analytical system.  This would mean if you were sampling an
aerosol with that kind of a bias associated with it, that you could
actually be looking at concentrations as low as 25 to 30 µg, but
interpreting those concentrations as being up around the PEL of 50 µg.

		The conclusion based on a review of these samplers is that these
samplers and respirable dust definition issues can and will adversely
impact results of anybody analyzing samples collected by the cyclone
samplers proposed by OSHA.

		OSHA has not demonstrated in the proposed rule and their analysis,
have demonstrated these samplers are actually feasible, excuse me, the
sampler, GK 2.69, and other high volume samplers, are actually feasible
in obtaining results that will allow for a fairly precise estimation of
the error and bias associated with this sampling such that you can
basically provide regulatory base samples.

		I had the single -- I had the opportunity of doing one other review
for the client, and that is a review of a specific paper that was
proposed or put into the PEA that dealt with a study done by Esswein,
looking at fracking operations and the exposures to quartz associated
with those operations.

		While this is an interesting study, it was purported by OSHA that this
was a baseline study of the industry.  And although I think that's a
kind of significant extension, it's not what the authors claim.  The
authors did not make such a claim, and I don't think that claim could be
made given that there was a small number of sites that were visited,
they were not selected randomly from a large population, and they were
basically sites that volunteered to be studied.	

		In this study there was a -- really, a relatively small number of
samples were collected from the monitored population.  There were like
15 different occupations that were available for sampling for quartz. 
Only 7 of those had four or more samples collected, and those four or
more samples were collected from basically 11 different collection sites
so that not every site had samples collected for every occupation they
demonstrated.

		And, in fact, almost half of the air samples were collected from one
occupation.  That was one occupation that was heavily exposed,
potentially heavily exposed; that's a sand-mover operate -- occupation,
excuse me.

		They also had issues of combining measurements from different sites,
resulting in what appeared to be population description parameters that
were -- I would say they were inaccurate and misleading.

		The problem being is that at one of them there was, in one particular
occupation, there was a geometric standard deviation -- a reported
geometric standard deviation of 7 to 8, and that is, actually in my
30-some-odd years, 35-some-odd years of working in this business, I have
never seen a GSD of quite that magnitude.

		Three to four is -- I think was discussed earlier as being an
uncontrolled environment, while two is being controlled.  And when I
have seen that kind of variation, when I've seen it, it has been
typically from a combination of combining various occupational groups,
sub-occupational groups into one occupation that really have different
levels of exposure.  And it distorts what you can obtain from the
information.

		During the study that was conducted by Esswein et al., they used this
sampler that I discussed earlier, GK 2.69, available from BGI
Incorporated.  

		And the thing that I found that would have been really helpful for the
study to understand their results, given that there was this bias
associated with the particular methodology they used, if they had
collected side-by-side samples, this would have been really, I think,
helpful in understanding, will this new cyclone, the high volume
cyclone, provide a sampling that will be effective for the industrial
hygienists in the field to then gather samples sufficiently large enough
they can be analyzed?

		And that ends my comments on the two areas of the OSHA PEA that I
looked at.

		JUDGE SOLOMON:  Next.

		MR. LIECKFIELD:  Hello.  My name is Bob Lieckfield, and I've got 37
years of analytical experience.  And the only point that I am testifying
to is the ability of the current analytical methods to measure
crystalline silica at the levels required under the proposed PEL.

		And with the 1.7 Dorr-Oliver, we're looking at levels of 10 µg at the
action level and 40 µg at the PEL.  If we use the high-flow sampler,
which Dr. Hall just talked about, we're at 25 µg for the action level
and 100 µg for the PEL.

		My position, at this point, is it's not technologically feasible to
measure at those levels.  There is literature.  Dr. Stacey from the
Health and Safety Executive did a paper in 2007, where he demonstrated
that the -- trying to measure at those levels would be fraught with
error.

		The accuracy that I'm speaking about is plus/minus 25 percent at those
concentration levels.  They're -- the x-ray method -- and I'll focus
on the x-ray method.  To summarize, there's x-ray, infrared, and the
colorimetric method.

		We won't talk about the colorimetric method.  It is rarely, if at all,
used.  In my experience the main method is x-ray diffraction, which
allows you a direct measurement of polymorphs, α-quartz, cristobalite,
and tridymite.  

		The proposed standard seems to imply, to me at least, that x-ray would
be the method of choice, because there is a standard for quartz,
cristobalite, and tridymite.  And I have some comments about how that
would work, through the measurement process.

		The infrared method is -- there is not a direct way to measure
α-quartz and cristobalite.  You're doing it by a mathematical
manipulation, which would be secondary to a direct measurement.  The
infrared method is also prone to interferences.  It's very good in coal
mines because it's a characterized dust.

		In general industry, the question would come up, what type of
interferences might an individual site expect, which then pushes the
technology more to x-ray diffraction.

		So we'll focus on the x-ray method.  The other -- the last point of
that is OSHA only has a method for x-ray.  They did not show an FTI
here, so that's why I'm focusing more on the x-ray method, but I don't
believe either method is capable of measuring accuracy at plus/minus 25
percent at the levels required of the action limit or the proposed PEL.

		To substantiate this, we look at the AIHA PAT samples, which is an
industry standard for performance testing.  They are more real-world
samples.  They give a better measurement of what an individual lab can
produce with a particular method.

		That's a better measure to look at than an internally generated
precision and accuracy.  If you take a pure standard, which is what OSHA
put in their PEA, spiked samples in the lab, known concentrations, you
can tighten up that accuracy, or it would appear that you can.

		So my comments are more in the real world when you're producing data. 
And I come from a production lab, so we're producing, you know, 50 to
100 data points a day for silica.  The industry requires or demands to
fit this, for employers and consultants, whoever is taking silica
measurements, that the number they're going to get is representative of
the sampled environment.

		But if we look at the PAT samples, the error at -- and the lowest
they have is 40 µg, so we're not at the concentration levels that we
still need to measure, but between -- greater than 40, the error bar
around there is plus/minus 30-some percent.

		Specific numbers would be at from 40 to 50 µg.  It's plus/minus 35
percent.  And this is two standard deviations, which is the OSHA
requirement for accuracy.  From 50 to 80 µg is plus/minus 32.  From 80
to 100 it's plus/minus 31, and 100 to 200 is plus/minus 29.

		I indicate that I probably should have said it in reverse, but if as
you -- as loadings go higher, your error starts to decrease. 
Similarly, as loadings go lower, your error is going to increase.  So
below 40, we don't know what the precision and accuracy of the
analytical methods are.

		We can surmise that it has to be somewhere in the neighborhood of 30
percent or more, which then does not hit the -- OSHA's precision
requirements for a valid analytical method.

		And not to -- I'm going to summarize the data that was in the PEA on
the OSHA labs PAT performance.  The numbers I just gave you was on
the -- excuse me, the community of industrial hygiene labs within the
AIHA PAT Program.

		But if we look specifically at the OSHA lab -- and this is not saying
anything that the OSHA lab does not employ the method.  This is all
about the method itself, not the people running it.  That they had 81
percent of their results were within plus/minus 25 percent of the
reference value; 38 percent were -- excuse me, let me rephrase.

		Forty-five percent of the data was below -- outside of that standard,
reference standard for AIHA.  And 65 percent was an overestimate -- 45
percent underestimate, 65 percent of the data would be overestimated,
when then puts things into -- for compliance testing, a bit dicey,
because there would be always some questions as to whether an employer
was in compliance or not.  

		And if -- for just general studies with industry, they would not have
a good idea of that either, given the methods that we're using.

		What I am proposing to OSHA before -- in consideration of the
analytical testing, is to do further studies on the methods that'll be
used for compliance testing.  Define the variability across the
laboratories.  What is the source?  How can we improve it?

		Dr. Eller of NIOSH did that back in 1999 and made recommendations. 
That was 14 years ago.  Probably most of those recommendations have been
incorporated.  I think we need to look at that again and hone in on
improving that method to meet that plus/minus 25 percent.

		The limits of detection and limits of quantitation, the lab industry
has various ways to go about doing that.  I think we need to define,
especially for silica, a set way of doing LOD and LOQ, because that is a
variable, and also collect data, assemble data in the area, in the range
that we're actually needing to measure, which would be somewhere between
10 and 50 µg of silica.  And that ends my comments.

		JUDGE SOLOMON:  Okay.  Who is next?

		DR. KNUTSON:  I'm Gerhard Knutson.  I'm the President and senior
consultant in a small consulting company, Knutson Ventilation out of
Minneapolis.  I welcome the opportunity to speak before OSHA and the
peer reviewers, especially since I get to come south in order to be able
to enjoy your winter as opposed to our winter.

		My testimony today is going to be focused in one small area.  I'm
going to be looking only at Appendix A.  Appendix A has to do -- the
Silica PEA has to do with hydraulic fracturing.

		And I'm even going to narrow it down a little bit further than that. 
I'm going to look at one item in that, and that is specifically I'm
looking at feasible engineering controls.  

		And the bottom line, the conclusion I draw, is that in Appendix A of
the Silica PEA, OSHA did not establish feasible engineering controls
capable of reaching or even approaching the proposed permissible
exposure limit.  And some of the things I'll follow will kind of
substantiate why I drew that conclusion.

		First of all, I think we have to back off and we have to ask what is
hydraulic fracturing?  And there's two aspects to it.  One of the
aspects is what goes on underground?  And this is the magic where you
have very high-pressure slurry that's got a proppant that forces the
shale to fracture.  It takes the fracture in the shale and opens it up
and puts a proppant in there to prop it open, so that natural gas can be
extracted from the well.

		We're not going to talk about that magic, although that sounds like an
awful interesting topic.  I'm only going to look at what is above
ground.  And the aboveground portion of it is the handling of the
proppant.

		And for today's talk, I'm only going to assume that we're looking at
crystalline and free silica as the proppant, sand.  And that's what most
people are using most of the times anyways.

		So when we look at it, we have to look at what goes on above ground. 
And the things that go on above ground is, the sand gets shipped in. 
It's not manufactured or it's not classified.  It's not -- nothing's
done with the sand at that hydraulic fracturing site.

		It gets brought in by car -- or trucks, rather, pneumatic conveying
trucks.  And the pneumatic conveying trucks then convey this sand over
to big hoppers.

		The big hoppers discharge down onto conveyor belts, which move over to
other conveyor belts, which move down to a hopper that feeds an auger,
that feeds a mixer.  

		And the mixer now takes the proppant, adds water to it, adds the magic
that goes into the mystery that really happens underground, some
chemicals that are added in order to make the process work better.  And
then it goes out in a sealed pipe.  Once it leaves the mixer, there's no
realistic exposure to silica at all.  

		So the question is, where does the exposure occur and what has to be
done about it?  Well, NIOSH, in its review of several hydraulic
fracturing sites, came up with seven different sources of airborne dust,
airborne silica.

		Remember, this proppant is almost all silica, so when we talk about
dust and we talk about silica and we talk about proppant, we're talking
about the same thing.  So it -- not always respirable, because some of
this stuff is pretty big.  But at any rate, what we have to worry about
is where these sources are.

		I don't disagree with the sources that NIOSH identified.  I might have
gone a little bit differently and taken some of their sources and broken
them down to multiple sources within the source, but they generally
found the sources that were critical.

		In order to understand hydraulic fracturing, I think you have to go a
little bit further than just thinking about conveyors.  Let's look at
the characteristic of the material that's moved.

		The sand that gets brought in to the operation has been worked on
before it gets there.  Specifically, it comes out the quarry, it gets
crushed.  It gets washed.  It goes over to a dryer so that it becomes
bone dry.  And then it goes through a classification process that gets
the particle size of the proppant into a very narrow band.

		And there's multiple bands of proppant that are used in the hydraulic
fracturing operation because of the magic that goes on underground.  But
it's very critical that you don't mix them.

		So now we take this proppant, and we put it into big hoppers.  Since
in a given well site you may end up using one proppant in the beginning
and one proppant in the end, you've got different hoppers for different
size materials.  And now you just charge that on.

		But I think we also have to sit back and say how big is this
operation?  And people in the hydraulic fracturing have a very nice unit
that they use.  They talk about sacks.  A sack, a 1 cubic foot, a
100-pound sack of sand.  They don't have any sacks but they talk about
sacks.

		And when you look at hydraulic fracturing, you're talking about
production rates in the order of 200 sacks per minute.  Two 100-pound
sacks, 20,000 pounds, 10 tons per minute, 600 tons per hour; this is a
big operation.  You're moving an awful lot of material.  We'll come back
to that.

		In the PEA, OSHA looked at several different research papers that have
been published and tried to use those to show that it is technologically
feasible to meet the standard.  And I'm going to look at two of those in
particular.

		One was a study -- and both of these are applied, not to the most
obvious source, because that's the transport of the sand through
pneumatic conveying, but through the sand handling at the sand movers,
at the conveyor belts, and at the hopper feeding the mixer.

		And in there, they looked at one study that was a stone-crushing
operation in Iran.  Let's think a little bit about what this operation
was and what happened.  Rocks were manually charged into a crusher, that
went into a grinder, that went down into a conveyor system that went
over to a series of screens, that got bagged out and shipped off to be
processed by some other plant somewhere else.

		The materials that were used were high grade silica -- 97 to 100
percent silica, and low grade, which was 80 percent silica, still pretty
high silica content.  The operation, on a daily basis, typically had 6
to 8 tons per day.  Compare that to hydraulic fracturing, that's one
minute of operation, corresponds to their daily production.  

		The approach that was used, I fully support.  The idea was to take a
look at these operations -- many of them are typical operations, at the
crushing facility, go to the industrial ventilation manual, look up the
ventilation sketches, the VS prints, see what they suggest, apply those
prints to the ventilation system and make it work. 

		And that's precisely what this research team did.  I've had
correspondence with them, and they've identified which VS prints they
used.  It's in my report.  Many of the prints were for canopies,
canopies over the crusher, for enclosures around screens.  So these are
listed, and now the system was put into effect. 

		And another paper, not the one that was cited by OSHA but a second
paper by the same authors that talked about the other end of the
ventilation system, the air pollution aspects of it, talked about
volumetric flows.  You can go back and look at the flows for it.

		And a moderate flow is 15,000, 20,000 ACFM type of operations, not
horrendously large but a fair amount of air.  And so when we look at
what happened here, we had a marked reduction in exposures.  Well, you
start out with a rather primitive operation that doesn't have any
control in it, and what happens?  Reasonable controls give us very good
reductions, in the neighborhood of 99 percent reduction.

		I mean, that's a phenomenal reduction.  On the other hand, when you
look at it, the reduction that was observed in this operation did not
bring the exposures into compliance with the current permissible
exposure limit.

		Not only did they not do that, but when you go -- and this was a
factor of two to four -- well, about two times the amount of the
current exposure limit, but if you look at where the operators are in
the crushing operation versus where the operators are in hydraulic
fracturing, they're in different locations.

		In the crushing operation, the operators were moving material but not
spending a whole lot of time next to a crusher.  I've been in a lot of
crushing operations.  Nobody in their right mind, except an engineer
trying to figure out how to control it, is going to stand near the
crusher.

		And that's where the real high exposures -- in the hydraulic
fracturing, remember, we're moving 600 tons an hour, 10 tons a minute,
300 pounds a second.  If you aren't watching what happens, you overflow
the bin, you spill on the ground, and we're not talking about a couple
of pounds.  We're talking 300 pounds a second falling on the ground.

		You can't not be there in order to make sure that operation exists. 
So even though the operator exposures were only twice the permissible
exposure limit, at this marked -- the system did a great job of
reducing exposures.

		When you look at the area samples that are more representative of the
kinds of exposures that hydraulic fracturing operators would have if
they were in there, now you're talking about 100 times.  That's a
substantial excess of the permissible exposure limit.	

		You're not talking fractions of a milligram; you're talking about
milligrams.  So it's a very substantial exposure that existed at the
stations, which is more consistent with what would have happened had
this been similar to hydraulic fracturing.

		The problem is, none of the equipment matched.  None of the controls
that were used are applicable for hydraulic fracturing operations. 
Still, the exposures were very high.  And, obviously, the production
rate was miniscule.

		The second operation that was cited by OSHA is also a crushing
operation.  This is stone crushing, again, and now the rocks that are
coming in are a little different grade.  They're lower silica, and the
silica samples that were reported in the study were anywhere from 0 to
27 percent of both quartz and cristobalite.  The 27 percent was the sum
of the two.

		And so the quantity of crystalline silica in the rock material was
less.  The approach that was used -- well, let's go back and talk about
the operation.  The operation manually charged these rocks into
crushers, brought it down to a grinder, and then took the material that
came out of that, and that was their end product.

		They have a beautiful picture in the report that really explains
what's going on.  They have a woman with a basket of rocks on her head
walking up to the crusher, another woman standing at the crusher with a
basket of rocks waiting to dump it into the crusher, and another woman
walking out of what appears to be a tunnel -- appears, because there's
so much dust in the air you can't see what's going on, that would have
been directly under the crusher, again with a basket on her head full of
crystalline silica.

		Now, I don't know how many sacks per hour or per minute this
corresponds to when you've got people carrying them on head baskets, but
I guarantee you it's not anywhere near 200 sacks per minute, probably
closer, at best, to 1 or 2 sacks per minute, because one of the pictures
showed several men and children -- because child labor was a problem in
the area, that were carrying this material and hand dumping it in there.

		How this can be related to hydraulic fracturing that is mechanically
moving 20 sacks per minute through a very carefully controlled system,
I'm at a loss.  The production rates would have been a fraction of it. 
The equipment that was used was completely a disjoint from any of the
equipment that is used in hydraulic fracturing.

		The process that was used was misting and using of water.  And the
study wasn't clear on one other thing.  It used water techniques, but it
didn't talk about exactly what it did.  There was a picture of a water
spray in a chute that looked like it was capturing the material coming
out of the crusher and wetting it.

		And you could see an accumulation of mud on the side of the chute. 
They also did some misting.  Well, go back to hydraulic fracturing. 
What have we got?  We have sand that has been dried, classified, sized
very carefully. 

		And if you add water to that, if you wet the material, you are really
in trouble, because the material in hydraulic fracturing, the media that
is being used, the material itself, almost flows like water.

		It's dry, it's crystalline, and the hopper bottoms aren't real steep. 
You don't have to have a 60-degree hopper bottom.  You can get by with a
45-degree hopper bottom, because it flows so well.  Add a little bit of
water to it, and now you have to go inside and muck it out in order to
get the material to move.

		It sticks onto to the belt, it sticks in the hopper, it screws up the
augers.  So you can't wet the material.  You can use misting in the air.
 And the concept that misting in the air is that you will take the small
particles, and it meets the larger particles, and through Brownian
motion or impaction or whatever mechanism, the particles get to be
bigger.

		Now they're bigger and they fall out.  If you take a small, respirable
particle and you engulf it in a droplet of water, now it's no longer
respirable.  That makes a big difference.  So it does have some
advantage.

		In the conclusion that they had, they didn't differentiate between the
benefit of wetting the material and the benefit of the misting.  And
that's a major flaw in the application of this.  Again, when you look at
the end result, the end result was exposures above the permissible
exposure limit, both the current and the proposed.

		So we've got a situation now that we have two controls, questionably
applicable.  Assuming that they are applicable and pushing the number
through, now what OSHA does is it takes the two controls and applies it
to the same operation.  Not a bad thing to do.  Obviously it's going to
improve the control.

		It should be better than both of it, but there is a logical and
scientifically incorrect assumption made, and that assumption is that
they're independent.

		Any of you that have anything to do with probability recognize what an
independent event is.  The problem is these two controls are not
independent.  They control the same error at some times, and you end up
double dipping.  

		And you have -- what you're essentially doing is you are taking
credit for the dust that is controlled by the local exhaust ventilation
and the dust that's controlled by the misting simultaneously.  And that
doesn't happen.

		And in my report I made up a problem and put some numbers on it, ran
some numbers, and showed how doing it one way gives you one number and
the other one gives you about a 30 percent difference in the efficiency.

		The approach that was used by OSHA, by assuming the independence of
control efficiency, is wrong.  One minus the efficiency is not equal to
the product of the quantities one minus the efficiencies.  It's going to
be much less than that.  And the more product items you have in there,
the greater the difference.

		So where do we stand?  We have a situation where I think if we had a
bunch of experts get together and discuss what does it mean to have to
show the requirement to demonstrate feasible engineering controls to
meet the permissible exposure limit or meet a proposed permissible
exposure limit, you'd have some -- quite a bit of disagreement as to
what all that means.

		But I think you would have unaniminity that it's not a numbers game. 
We're not playing with numbers, that you really have to sit back and
look at it.  I think you'd get agreement that it is not the
juxtaposition of control measures that did not work, in industries that
are unrelated to hydraulic fracturing, with equipment that's unrelated
to hydraulic fracturing, and draw any conclusion about feasibility in
hydraulic fracturing.

		And as you can see, I firmly believe they did not establish that there
are feasible engineering controls that either are known or can be
reasonably expected to achieve the proposed permissible exposure limit.

		JUDGE SOLOMON:  So, Dr. Knutson, we're at about 3:05.  How much time?

		DR. KNUTSON:  Zero.

		JUDGE SOLOMON:  Pardon?

		DR. KNUTSON:  That was my last statement.

		JUDGE SOLOMON:  Oh.

		DR. KNUTSON:  Your timing is excellent.

		JUDGE SOLOMON:  You have one more witness?

		MR. CHAJET:  We have one more witness, Dr. Bunn.

		JUDGE SOLOMON:  So we'll play a little musical chairs here.  If you'll
move forward.  State your name, please.

		DR. BUNN:  My name is Dr. William Bunn.  I've served as medical
director for multiple industries from 1982 to 2013.  That included
Bristol-Myers, Manville, ExxonMobil, and Navistar International. 

		My responsibility during this period of time for over 100,000 people,
about a third of those were exposed to significant levels of crystalline
silica.  There were many exposures above the PEL in these groups;
however, with effective programs, I never saw or was advised of a single
case of silica-related disease during my tenure as medical directors.

		I provide comments today on behalf of the U.S. Chamber of Commerce in
an effort to help focus the silica rulemaking on the reality of
conditions in the U.S. workplaces, separate from statistical modeling
and other approaches.

		In my experience, I do not believe that we necessarily need to change
the rules.  My interest in lung disease and lung cancer began with my
fellowship, sponsored actually by the National Cancer Society and NIOSH
at Duke University and University of North Carolina.

		I was privileged to see a number of folks with silicosis during that
period of time and treat them, so it's very -- I recognize it's a very
significant disease, and took care of a lot of folks.

		My first role as a medical director, having previously served as a
consulting director, actually, for EPA and IEHS, was a large corporation
where we dealt with factory ceramic fibers, roofing, mining, paper
products, forest products, all were areas where there was significant
amount of crystalline silica exposure.

		Of note were the mining operations, and this corporate medical
director role included diatomaceous earth, milling, mining, and
calcining.  And diatomaceous earth needs a small amount of crystalline
silica when we start it at 1 to 5 percent.

		However, as we go through the milling and the heating of the
substance, we come out with primarily cristobalite.  It starts as
quartz, goes to cristobalite.  It's mainly used for things you'd
recognize.  It's actually filtration in swimming pools, filtration for
pharmaceuticals and beverages and other things.

		While I was there, I was privileged to be a part of a Harvey Checkoway
study.  And you're going to laugh at me.  This was at a plant that's
been studied for many, many years in California.  Many questions have
risen with unions and management.

		That study finished up in the late '80s.  It was updated in the '90s,
and we have data past 2000.  I talked to Dr. Checkoway.  We put
together a table, which I've included in my comments.  I would ask that
you take a quick look at it.  And it comes from his study of mortality
among workers in the diatomaceous earth industry.  This is a study
that's been significant in IARC and other findings.

		Basically, what we found, as we look back, is that for folks who
entered the cohort after the 1960s, we didn't have any further cases of
silicosis.  So if you look at the table, we had large numbers of
silicosis in other cohorts of workers.  When we got to the '60s, we
didn't have further cases.  And we followed these cases out to past
2000.

		So what that suggests is something happened.  We know what happened,
is we changed the ventilation, we enforced protection, respiratory
protection, and we put in great programs in cooperation with the unions
there.

		In the same role, I was in charge of fiberglass operations. 
Fiberglass is obviously made from sand, so we had significant exposures
to folks bringing in the sand and folks putting it there.  I never saw a
case of silicosis in this area.  Because of the concerns about fibers,
we actually saw every patient, did PFTs, three-year chest x-rays on
every single individual in a fiberglass plant.  So these people were
very closely monitored.

		Forest products may sound interesting, but I was -- you know, people
who are working with trees, there's a lot -- particularly in the south,
where we exceeded the PEL on a number of cases and we used masks, and
that's somewhat surprising.

		My next role was a petrochemical company.  We didn't really see --
this was really before fracking, but we didn't see anything even though
we did have some levels close to the PEL.  We used personal protective
equipment during that time.

		My final employer of the last 18 years was Navistar.  Our primary
exposure was in foundries.  This is where the engine parts were made. 
Basically what happens is you build a mold.  The mold is built with
crystalline silica and other components.  You pour the molten metal into
the mold.  Then the challenge comes is to get -- after it cools, is to
get the mold away from the metal part.

		This is very difficult.  The chipping part is something we really have
to do by hand.  You have shaking, but there's really a hands-on piece of
that.  What we determined over the years is that we couldn't meet the
PEL in most cases.  We could get very close.

		To give you an example, it took about a 60-mile-an-hour face wind in
ventilation in the area to get close to the -- so we used a combination
of ventilation and personal protective equipment.  We monitored
everything closely.  We worked with our unions there.  And in over 18
years we saw no cases of silicosis and no workers' compensation cases.

		So what can I say?  I can say that over a 30-year career, responsible
for 100,000 employees, many exposed well above the PEL, I saw no new
cases of silicosis and none reported or were litigated.

		I think one of the common protection was a good combination of
engineering controls and personal protective equipment.  I would raise
the point that to put together good protective programs, we need to
consider protective equipment, respiratory protection, other changes,
wetting.

		It's a critical component is this respiratory issue.  We've looked at
it over years.  It's improved a lot.  We have much better respiratory
protection than we had in the past.  And we really haven't had problems
with making that a part of the process.  It's not the only answer, but
many cases, it's the part of the answer that we have to have, for
fracking and for other things.

		So everything's improving, and I think that's important.  Thus, I
think it's important that OSHA consider PPA as primary control,
establish a focus on education in workforce for the approximately large
number of employees that actually exceed the PEL.

		We haven't seen a particular evidence of anything of -- in our plant,
with effective programs, we've not seen silicosis.  We've looked some,
at actually, at the cancers as well.  So I guess the bottom line is that
in my 30 years of experience using engineering and personal protection
controls in good cooperation with the workforce, silicosis can be
prevented.

		JUDGE SOLOMON:  Is that your presentation?

		MR. CHAJET:  It is, Your Honor, for today.

		JUDGE SOLOMON:  The peer reviewers, are any of you -- do any of you
want to ask questions?  Dr. Ginsberg.  Okay, now while he is going up,
in the audience, are there any of the public who want to ask questions? 
There are a number of hands.  Let's start at the first row which is over
here to my right, your left.  You'll be the next person.  You want to
get in line?

		MR. CHAJET:  Your Honor, I do want to move into the record the vials
of silica.  And I want to put a photograph up on the screen.

		JUDGE SOLOMON:  Okay.  So let's do this by the numbers.  Ms. Ryder?

		MS. RYDER:  Sure.

		JUDGE SOLOMON:  You have those exhibits and the --

		MS. RYDER:  Not that I know of.  Were they in the PowerPoint
presentation?

		JUDGE SOLOMON:  There are the vials right there, and --

		MS. RYDER:  Oh, these.

		JUDGE SOLOMON:  -- I asked if there was an envelope.  If you have an
envelope, maybe you can mark the envelope, put the vials in the
envelope.

		MS. RYDER:  Okay.

		JUDGE SOLOMON:  I mean, this is a public hearing.  It's not like the
documents are going to be sealed or anything, but --

		MS. RYDER:  Do you want to identify these?

		JUDGE SOLOMON:  Yes.  That's a good idea.

		MR. CHAJET:  Yes.  Those are three sealed vials or bottles that
contain the amount of silica or dust that would be needed to fill the
air in a room approximately 20 by 30 by 20, a couple of different vials,
depending on what the PEL is.

		Here you've got the material in teaspoons, demonstrating it.  We
wanted to put that in the record to show the amount of silica that is
the equivalent of the PEL in a large room, as compared to the railcars
full of silica sand or the silica sand that you see on the beach.

		JUDGE SOLOMON:  There are three vials.  How do you differentiate among
the vials?

		MR. CHAJET:  Let me ask Bob to identify them.

		JUDGE SOLOMON:  Is there some way to mark them so that we can know?

		MS. RYDER:  Yes.  They're labeled.

		UNIDENTIFIED SPEAKER:  They are labeled.

		MR. CHAJET:  Yes, they're marked.

		MS. RYDER:  Just to clarify, is this respirable crystalline silica?

		UNIDENTIFIED SPEAKER:  Yes, it is.

		MS. RYDER:  Okay.  All right, so I think these will be marked as
Exhibit 8 --

		JUDGE SOLOMON:  You want to read what's on them?

		MS. RYDER:  -- for 5.6, nine, Exhibit 9 for 12.7, and Exhibit 10 for
25.5.

		JUDGE SOLOMON:  Okay.  And there's no objection, so those are admitted
into evidence.  

(Whereupon, the objects referred to as Hearing Exhibits 8 through 10
were marked and received in evidence.)

		JUDGE SOLOMON:  Each one of those, I guess, will get an envelope that
will be marked, is that --

		MS. RYDER:  Yes.  And they'll be available for inspection in the
docket office.

		MR. CHAJET:  And we'll give you a copy of the slide.

		JUDGE SOLOMON:  Okay.  Now, to the photograph, what is the photograph?

		MR. CHAJET:  The photograph is the amount of silica reflected in those
vials, but on a teaspoon measuring device.

		JUDGE SOLOMON:  Do you have a copy for the record?

		MR. CHAJET:  Not with me, but I will provide one.

		MS. RYDER:  Okay.  I will reserve Hearing Exhibit 11 for that
document.

		JUDGE SOLOMON:  Okay.  Dr. Ginsberg, you have a question.

		DR. GINSBERG:  Yes.  I have two questions for -- is it Dr. Bunn?

		DR. BUNN:  Yes.

		DR. GINSBERG:  Sorry.  I didn't catch your credentials.  And one
question for Mr. Hall; is that right?

		UNIDENTIFIED SPEAKER:  Dr. Hall.

		DR. GINSBERG:  Dr., I'm sorry.

		DR. HALL:  That's okay.

		DR. GINSBERG:  I didn't pay close enough attention during the intros. 
All right, so Dr. Bunn, we heard something about exceedances earlier
today, about how they might explain current problems with silicosis that
you wouldn't expect at the PEL, but exceedances above the PEL.

		So without knowing really the industrial workplace like you do, when
there is an exceedance at say your facility, and OSHA comes in and finds
it and maybe gives a fine or a warning or whatever they do, in 30 --
and that was supposedly in about 30 percent of the cases across a
variety of industries, we saw some exceedances -- what happened?  How
quickly are those exceedances fixed, and is that, tend to be an ongoing
problem that would lead to a long-term exposure above 0.1?  Or when OSHA
finds a problem, does that tend to get addressed, say at your
facilities?

		And then regarding the PPE question, I'm not sure exactly why, maybe
you could educate us, why wouldn't -- why hasn't OSHA historically used
PPE rather than engineering controls to regulate these kinds of
exposures?  And what might be the issues in a facility that isn't as
tightly managed, medically managed, and --

		JUDGE SOLOMON:  So, are these for Dr. Bunn or for both?

		DR. GINSBERG:  Yes, these two are.  What are the issues with PPE and
what -- and how well do these inspections then get you down to the PEL?
 And then the question for Dr. Hall is, well, we have an issue with
needing to get more dust on a sampler so you can accurately measure
quartz.  

		So you don't want -- so you want to have, on the one hand, the high
flow, the high volume, but you're saying that there's an inaccuracy
there, a potential positive bias, so what do you recommend OSHA do to
get more sensitivity going forward?

		JUDGE SOLOMON:  Okay.  I'll start with Dr. Bunn.

		DR. BUNN:  Yes.  The first question on OSHA inspections is that
generally we do regular -- OSHA notwithstanding, and as OSHA comes in,
we usually have crystalline silica measurements.

		In areas where we have difficulty, we actually sit down with OSHA and
reach an agreement, which really key to sort of three questions.  One is
that, what can we do with engineering controls?  At our foundries we
really, there's just literally -- and OSHA agrees, there's no feasible
way to get there except for the use of respiratory protection.

		And I would emphasize, this isn't the whole plant.  This is at certain
specific, restricted areas, and they could be easily observed.

		So generally -- I'm trying to -- generally we haven't had fines. 
What we've had is meetings to sit down and work together.  And I think
that's worked out quite well.  So that's -- so the easy answer is we're
working with OSHA all the time because we know we have areas are going
to be high, and we're working with them to show them that we've done the
best thing we can.

		And it's sometimes a challenge, because engineering equipment breaks
down, too.  And people may not wear their masks, people with -- so
we've really got to have a good inspection and maintenance system and
convince them of that.

		So the answer is, when we have an exceedance, we fix it if it's at all
possible, and if not, we work with OSHA.  And NIOSH has been in a lot of
our plants, too, like in Franklin, where we've had NIOSH in to help us
out, in a cooperative effort to get to the lowest possible level, and
then to use respiratory protection along with those.

		Question -- so that's what happens in our plants.  And there's always
a question, but the question always is well, that's a big -- well,
we've purchased a lot of small companies.  Some have had 20 people. 
And, you know, this kind of approach could be used, I would say, in most
facilities.

		The second question is a tough one for me.  You know, as a physician
we wear a mask in operating rooms, and people who have infectious
disease.  I worked in the pharmaceutical industry.  We had, could wear
them with animals, because of allergies and other issues.

		Yes, actually I worked at EPA for a while, and we had this -- so I'm
very, you know, I'm used to wearing a mask in a lot of different roles. 
I think there is a role for us, for OSHA to sit down and look at it.

		I mean, we have primary preference for engineering controls.  I don't
know if we want to change that, but what I do think is we really don't
give much credit for respiratory controls, the way things are written
now.  We really ought to look at an integrated approach that -- there's
sort of three pieces to it.

		There's engineering, there's respiratory controls, and there's other
safety controls and prudent cleaning.  There's a lot of other things we
do in our plants to make sure we keep the operation clean.

		So that's -- you want my -- my easy answer is, I don't really know
why we've not used them.  Perhaps and -- if we look back, there's some
bad examples.  You know, it's 40 years of respiratory improvement. 
That's an important issue.  It's a lot better than it used to be.  It
used to be less comfortable.

		You know, I think there are employers who try to use respiratory
control, perhaps, instead of looking at engineering.  Ventilation
systems have changed a lot, too.  So they're always there now, pretty
much.  You know, we look at construction, whatever, but in our industry.

		So I guess the answer is yes, I think now may be the time to put more
respiratory protection into the approach.

		JUDGE SOLOMON:  Dr. Hall, you remember the question?

		DR. HALL:  Barely, but I remember -- I would like to comment further
on what Dr. Bunn said about --

		JUDGE SOLOMON:  Why don't you answer the question first, and then
we'll go on.

		DR. HALL:  What would I expect OSHA to do, specifically, about the
biases associated with the high volume samplers?

		DR. GINSBERG:  No.  I'm sorry.  What would you want them to do, given
that they need to collect more dust to be more sensitive, yet they --
you talk about this bias.  So what would you want them to do to fix the
problem?

		DR. HALL:  First up, they need to collect more dust.  Provided they
change the PEL and the AL to a lower level of 50 and 25 µg, they need
to collect more dust to reach a level of quartz. 

		And still, reaching the action level, even with high volume samplers,
is going to be dicey at best.  But what they could do is, specifically,
is characterize high volume samplers much more effectively.

		They haven't really characterized the GK 2.69, although they
recommended this, specifically looking at or making a requirement to
look at the environment you're going to use that sampler in, and to know
what the characteristics of the aerosol are, so that you can understand
the biases associated with the data that you grab.

		JUDGE SOLOMON:  Anything else, Dr. Ginsberg?

		DR. GINSBERG:  Yes.  I just wonder if part of the logic of lowering
the PEL is that if we acknowledge that there's always going to be
exceedances, sort of as a halo above the PEL, of -- you know, whether
they're temporary or whether they're systematic at one plant or another,
if you're lowering the target, you know, you always have, you know, the
bull's-eye and the halo effect.  That your exceedances then would also
shift downward so that you've got less problem at the high end.  Is
there a logic to that?

		DR. BUNN:  Well, I'll answer, part from a -- I work, currently, with
four plants that crush rock.  And the answer is kind of maybe, maybe
not, because they're -- when they have exceedances, and they have them
on a continuous basis, really, they have them when the equipment breaks
down because of the material they're handling.

		They only occur in certain parts of the crushing operation.  Now, at
this facility, they're all wearing respiratory protection whenever the
plant is operating.  So, and they're wearing respiratory protection that
meets the requirements of reducing their exposures.  And they have
little or no complaints.  These are union shops, too.  They have little
or no complaints from the folks.  

		I think, in my career, respiratory protection devices have changed
significantly, while we have, as a professional organization and
occupation, haven't really looked at how those changes have impacted how
we make recommendations and why we make the recommendations for control
that we make.

		JUDGE SOLOMON:  Thank you, Dr. Ginsberg.  You want to come forward
please?  State your name for the record.

		MS. TRAHAN:  Hi.  Chris Trahan with the Building Trades.

		JUDGE SOLOMON:  Can you spell your last name?

		MS. TRAHAN:  T-r-a-h-a-n.  Actually, I have a question for
Dr. Lieckfield.  Did I say your name right?

		MR. LIECKFIELD:  Close enough.

		MS. TRAHAN:  Sorry.

		UNIDENTIFIED SPEAKER:  Close enough.

		MR. LIECKFIELD:  I'll answer.

		MS. TRAHAN:  You said something that I found interesting.  You said
that the lab -- and correct me if I'm misstating what you said, the lab
industry has a way of defining LOD and LOQ -- the limit of detection
and limit of quantitation.

		MR. LIECKFIELD:  Yes.

		MS. TRAHAN:  But you didn't say what your industry -- the lab
industry, how you do that.  Can you explain that?

		MR. LIECKFIELD:  The point is, there is no standard way to define
LOD/LOQ across the industry.  Some use -- now I'm talking about the lab
industry, not individual labs.

		MS. TRAHAN:  Okay.

		MR. LIECKFIELD:  Some use a signal-to-noise for an LOD, and then 10
times signal-to-noise for an LOQ.  Other labs will use a low calibration
standard, something they can see --

		MS. TRAHAN:  Okay.

		MR. LIECKFIELD:  -- that they will call an LOD, but it's not
necessarily the analytical method LOD.

		MS. TRAHAN:  Now, how does your lab define LOQ?

		MR. LIECKFIELD:  We define it, for our routine work, as low standard,
something we can see.

		MS. TRAHAN:  So, is there accuracy and precision?  What is the
accuracy and precision for silica analysis at the LOQ?

		MR. LIECKFIELD:  Well, I'll answer the LOD question; it's plus/minus
50 percent, is what we would expect as a pass/fail at the LOD.

		MS. TRAHAN:  So what about the LOQ, because that's a higher level
obviously?

		MR. LIECKFIELD:  My whole point is, getting the industry to make sure
we're consistent in how we measure that.

		MS. TRAHAN:  Okay.

		MR. LIECKFIELD:  We do not measure an LOQ, we being Bureau Veritas, we
do not use the concept of LOQ.  We sort of mix it together, low
standard.

		MS. TRAHAN:  For a reporting limit?

		MR. LIECKFIELD:  For a reporting limit.

		MS. TRAHAN:  Now, how would you define a reporting limit, then?

		MR. LIECKFIELD:  It goes along with getting the industry to define all
these terms.

		MS. TRAHAN:  So it may be inconsistent from lab to lab?  How does your
lab --

		MR. LIECKFIELD:  It is inconsistent from lab to lab, yes.

		MS. TRAHAN:  Can you explain how your lab does that?  I'm just trying
to figure out if, for example, your lab has accuracy and precision at
your reported reporting limit.  Did I --

		MR. LIECKFIELD:  At the reporting limit -- and we use 5 µg as a
reporting limit, something we can see, with a low standard, 50 percent
of the time, meaning it's -- it can go from 5 to 7½ and down below 5.

		MS. TRAHAN:  Okay.

		MR. LIECKFIELD:  That is our reporting limit.  That is neither an LOD
nor an LOQ.

		JUDGE SOLOMON:  Since we have a short break, will the people who had
their hands up in the next row please get in line.  Go ahead.

		(Off microphone conversation.)

		MS. TRAHAN:  And one other thing that I think you said early in your
comments was that you had to be able to accurately measure 10 µg in a
sample if you're using a 1.7 L/min flow rate, in order to accurately
measure at the action level.  Is that correct?

		MR. LIECKFIELD:  Correct.  That's correct.

		MS. TRAHAN:  Okay.  Thank you.

		JUDGE SOLOMON:  Okay, next.

		(Off microphone conversation.)

		JUDGE SOLOMON:  We know you, but you have to state your name, and
spell your last name.

		MR. HEARL:  Hi.  I'm Frank Hearl, and I'm the Chief of Staff for NIOSH
here in Washington, D.C.

		JUDGE SOLOMON:  Spell your last name for us.

		MR. HEARL:  H-e-a-r-l.

		JUDGE SOLOMON:  Thank you.

		MR. HEARL:  Okay, thank you.  Yes, I have some follow-up questions for
Dr. Lieckfield.  Actually, she was kind of hitting on some of my
points.  I'd like to start out, though, with just a slightly different
set of things.

		At Bureau Veritas, you do quite a bit of silica analysis, and I was
wondering, when you do -- when you get a sample in your lab, do you
always examine those samples for oversize particles?

		MR. LIECKFIELD:  No, we don't.

		MR. HEARL:  You don't?

		MR. LIECKFIELD:  No.  And I should clarify, he's a doctor, he's a
doctor, he's a doctor.  I'm not a doctor.

		MR. HEARL:  Okay, got you.

		UNIDENTIFIED SPEAKER:  I thought he was.

		MR. HEARL:  Okay, so there is no check for the oversize particle
thing?

		MR. LIECKFIELD:  Correct.

		MR. HEARL:  Okay.

		MR. LIECKFIELD:  But we're also not doing compliance --

		MR. HEARL:  Right.

		MR. LIECKFIELD:  -- results.

		MR. HEARL:  Okay.  So -- and then, on this other point about the
sampling, normally a person would sample under -- using the current 10
mm Dorr-Oliver cyclone at 1.7 L/min for a full shift, 480 minutes, that
works out to be about 0.8 m3.

		So if you were to need to quantify something at 25 µg/m3, and you
collected 0.8 m3, would that not be 0.2 or 20 -- or 10, 20 µg, not 10?

		MR. LIECKFIELD:  If it's -- you have the same problem at 20 as you do
at 10.

		MR. HEARL:  But it's twice as much on the filter as what you were
saying you needed to get the 25.  Because 0.8 times 25 is 20.

		MR. LIECKFIELD:  Perhaps I misspoke.

		MR. HEARL:  Okay.

		MR. LIECKFIELD:  At the -- we're talking action level?  Or are we
talking --

		MR. HEARL:  Whatever level it is.  I mean, whatever --

		MR. LIECKFIELD:  I was speaking action level and PEL.

		MR. HEARL:  Okay.  I mean, at the PEL, it would be 50 µg times 0.8,
and that would give you the number of µg you'd need on the filter.  If
you're wanting to quantify 25 µg/m3, and you sample 0.8 m3, that should
be 20 µg.

		MR. LIECKFIELD:  Okay.  The math -- I will accept your math.

		MR. HEARL:  Okay.

		MR. LIECKFIELD:  But at the end of the day, the lab industry and the
current published methods do not have any data underneath 50 µg.  And
at 50, OSHA is plus/minus 26 percent, NIOSH is plus/minus 18 percent
without measuring bias, because accuracy is precision plus bias,
absolute value of the bias.

		JUDGE SOLOMON:  Dr. Schwartz, you want to move down?  People in the
next row who had your hands up, could you get -- move up?

		MR. HEARL:  One other thing, at least.  I went to the Bureau Veritas
site, and where it cites how you're going to supply services, like for
the new OSHA proposed PEL, and it says that the evaluation of exposure
limits for crystalline silica, the reporting limit, and then you've
got -- or LOQ -- in parentheses, actually on that, for quartz and
cristobalite is 5 µg per filter, and the limit for tridymite is 10 µg.

		So is that consistent with the way you describe reporting limit? 
Because you have LOQ in parentheses at your website.

		MR. LIECKFIELD:  There's a little bit of poetic license there.

		MR. HEARL:  Okay.

		MR. LIECKFIELD:  But at the same time, what we can measure -- there's
no -- I have no issues with the ability of the current methods to
measure down at the 5 and 10 µg level.  That's really not at all my
point.

		The point is, you cannot measure it accurately enough to meet the OSHA
requirements of plus/minus 25 percent.  It's always been that way.  I
mean, it's -- there's always been that variability.  All analytical
methods, as you know, have increased variability down near the LOQ.

		MR. HEARL:  Sure.  That's -- I mean, that's the reason why they also
give extra credit, basically, in terms of how far you have to exceed the
limit before they cite.

		MR. LIECKFIELD:  Right.  But my old position, from the commercial
industry side, we are not doing compliance --

		MR. HEARL:  Okay.

		MR. LIECKFIELD: -- or we're not providing analytical data for
compliance sampling.  We know what the variability or the overall method
precision is, and we can accept that.  When you get into compliance, I
believe OSHA themselves say that it has to be plus/minus 25 percent.

		There is no data to support that under 50 µg that I've seen.

		MR. HEARL:  Okay.  Thank you, Your Honor.

		MS. KEY-SCHWARTZ:  I'm Rosa Key-Schwartz, from NIOSH in Cincinnati.  I
have--

		JUDGE SOLOMON:  I have to -- I know how to spell your name, but the
Court Reporter may not.

		MS. KEY-SCHWARTZ:  K-e-y, hyphen, 

S-c-h-w-a-r-t-z.  I have three questions for analysis.  The PAT Program
provides analytical interferences for the laboratories to prove their
ability to analyze in the presence of mineral or interferences; however,
would you please speak to the article published in the Synergist in 2000
by Edwards, which states that proficiency testing data should not be
used for assessing the performance of an individual analytical method?

		MR. LIECKFIELD:  I can say I just disagree with that statement.

		MS. KEY-SCHWARTZ:  All right.

		MR. LIECKFIELD:  Because -- that was simple, Rosa, data coming from a
laboratory, we all -- the lab industry generates their own performance
data by making pure spikes, which will tell -- give you probably the
optimal precision and accuracy.

		The PAT Program adds a bit of mystery to it because you don't know
what your -- you know their performance samples, but you don't know
what concentration, so it adds a little mystery to the laboratory staff.

		We're -- to me, that -- the PAT samples are real world.  It's just
as if you were getting a client sample, in our case.  And it's measuring
that performance.  So I think it is very key.

		I guess I would ask it back the other way, why did Dr. Edwards think
it wasn't appropriate?

		MS. KEY-SCHWARTZ:  Do you get to ask me a question?

		MR. LIECKFIELD:  No.

		MS. KEY-SCHWARTZ:  Well, I would just briefly say that in the NIOSH
methods, we have validation data for that method down to 20 µg, so I
want to speak to that, and then --

		JUDGE SOLOMON:  Do you have another question?

		MS. KEY-SCHWARTZ: -- and then move on to the next question, thank
you.  You talked about the performance having been studied by NIOSH, and
you suggested it be studied again, which I think is a good idea.  

		Then the colorimetric methods are known to be significantly more
variable, and I just wondered if you knew that, the fact that the
colorimetric methods have not been reported at all for PAT samples the
past few rounds, and so we expect the variability to significantly
decrease in the PAT data.

		MR. LIECKFIELD:  I don't know that, Rosa, specifically.  On the
colorimetric labs, the last metric I looked when the PAT data was
assembled was three or labs were still using colorimetric.  I didn't --

		MS. KEY-SCHWARTZ:  Okay, they were -- I just wondered if you were
aware that they were not reported at all in the past few rounds --

		MR. LIECKFIELD:  No.

		MS. KEY-SCHWARTZ:  -- so we would expect to see a difference in
variability.  I think that's a very positive statement that you made
about wanting to study that again.  Thank you.

		MR. LIECKFIELD:  That's the purpose of --

		MS. KEY-SCHWARTZ:  Okay.

		MR. LIECKFIELD:  My whole position is that we need, as an industry, to
understand the analytical method, now at the levels that are -- would
be required.

		MS. KEY-SCHWARTZ:  All right.  Thank you.  And my last question is
that the National Institute of Standards and Technology has provided
standard reference materials that are deposited on filters down to 10
µg.  The quartz series is 2950 and the cristobalite series is 2960. 
Does your lab employ these calibration standards?

		MR. LIECKFIELD:  Not as reference standards.  We use standard
reference material to create the calibration standards.

		MS. KEY-SCHWARTZ:  All right.  Thank you.

		JUDGE SOLOMON:  Next.

		DR. SIVIN:  Dr. Darius Sivin, United Auto Workers Health and Safety
Department.  My office is in Washington, D.C.  Question is for
Dr. Bunn.

		JUDGE SOLOMON:  You have to spell your last name today.

		DR. SIVIN:  S-i-v-i-n.  Dr. Bunn, you stated that you're unaware of
any cases of silica-related disease in folks who entered the various
workplaces that you worked in since the 1960s; is that correct?

		DR. BUNN:  I started in the 1980s, but the 1960s was the --

		DR. SIVIN:  But the folks in the -- the folks you're referencing
started in the 1960s and later; is that correct?

		DR. BUNN:  Some of them did.  I'd have to look.

		DR. SIVIN:  Okay.  The reason I ask is that are you aware, in the
Checkoway study that you cited, there was a statistical increase in lung
cancer among folks entering the cohort in 1970 through 1979?

		DR. BUNN:  I have to go look carefully at that.  Harvey and I went
through this, and I didn't think we got that.  But I'll look at it.

		DR. SIVIN:  Are you further aware that there was a statistically
significant increase in nonmalignant respiratory disease, both in folks
who entered the cohort in 1970 through 1979, and in folks who entered
the cohort in 1980 through 1987?

		DR. BUNN:  We didn't look at nonmalignant respiratory disease as
opposed to fibrotic disease, obviously.  And, yes, I'll stop at that
area.

		DR. SIVIN:  Okay.  I'd like to ask you a few further questions about
your experience as medical director at these corporations.  It is well
known that many chronic occupational conditions don't show up on OSHA
logs or workers' compensation claims, and that many are misdiagnosed by
family physicians.

		Did you ever undertake a systematic analysis of your employers'
sickness and accident claims or health insurance claims for both active
and retired workers to determine whether silica-related or other
occupationally related conditions showed up there?

		DR. BUNN:  We did -- two things we do.  I was in charge of the health
plan also, and so we analyzed regularly for major trends.  So I'm not
going to say you couldn't miss an individual, but see, respiratory
disease was a major issue for us to look at, because we had about 70
percent smokers in our cohort most of that time period, which makes it
very hard to look at those other issues.  

		That's why I would say I have to go back and look at it because
smoking control was the key issue in the plants.

		We also -- as you probably know, we follow every employee till their
death, because we pay their healthcare claims.  So we do take a good
look at that.  And, you know, I can't -- as you know, we have meetings
regularly with the UAW.  We meet multiple times a year with the safety,
and we have a one-year with the senior management to project our goals
for the year.

		So I think mostly -- to make your point, we didn't see anything.  We
looked at different levels, but we always look because all of these had
crystalline silica or fiberglass, other exposures, very well.

		In a lot of the plants, we -- and we had health records, so we'd look
for trend analysis.  We did this -- I'm not sure if you were there yet.
 We actually had a question in one of our labor negotiations in
Indianapolis where we actually did a joint study, and it didn't show
anything.  So we looked at cancer.

		DR. SIVIN:  Foundry study?  Okay, I'm not --

		DR. BUNN:  It was in Indianapolis in the 2002 or '3 negotiations. 
Frank was involved.  We went back and looked at it.

		DR. SIVIN:  Okay.

		JUDGE SOLOMON:  Are there any other people who want to ask questions
from the public?

		DR. SIVIN:  I have a couple more questions for Dr. Bunn.  One, can
you tell us the average age of retirement and the average length of
employment of folks in silica-exposed jobs in the various places you
worked?

		DR. BUNN:  Gee, it's seven places, so go back.  For folks that are in
foundries -- well, there was a break in employment, so it's a little
more complicated for that.  Tend to be long employment periods of time. 
They didn't all work in the foundry, though, because, you know, there's
an engine plant.  They switched back and forth.

		DR. SIVIN:  I've been to both of them.

		DR. BUNN:  Yes, you -- and we've been there, so.  There was some
switching back and forth, but they were generally long-term workers. 
And they retired at anywhere from 52 to we had one at 81, so very, very
broad range.

		DR. SIVIN:  Do you now -- that's the range, but do you happen to know
the average?

		DR. BUNN:  No, I don't.  So I would guess it's at least 50 -- maybe
60.

		DR. SIVIN:  Okay.  I understand that folks are followed until death,
but in many cases they may have retired before they would have
silicosis.  Just one more question: could you describe the control
measures which you believe successfully prevented silica-related
diseases at your facilities?

		DR. BUNN:  Yes.  And then one thing I point out is, you know, when
people get to be 60 in our cohort, as you know, we don't ever lose in
the follow-up.  We still look -- we have disease management programs. 
We manage everybody with COPD, everybody, for example, so we get all the
smokers.  I mean, they're not lost to follow-up in our case.  You know,
there's also, as you know, death benefits.

		So, foundries, you know, are probably one of the most difficult
places.  The chipping operation, we tried bag houses, we tried face
velocity, came up with a combination of as much ventilation as we could
do, and respiratory protection.

		We had a joint agreement, I think, as we're part of it, too, with
OSHA.  NIOSH actually did a study there, and we did a combination
program, respiratory protection, restricted work areas, and as much
ventilation as we could get into that work area.

		That group we followed very closely, and really didn't see anything. 
And I'll be honest with you, we tried other ventilation options.  There
just wasn't much we could do.

		JUDGE SOLOMON:  Thank you, Dr. --

		DR. SIVIN:  Have you already entered the Checkoway --

		JUDGE SOLOMON:  I thought that was your last question.

		DR. SIVIN:  Okay.  This is not really a question.  Have you already
entered the Checkoway study into the record?

		DR. BUNN:  No.

		DR. SIVIN:  Then I'd like to enter the Checkoway study into the
record.

		JUDGE SOLOMON:  Would you hand it to Ms. Ryder?

		MS. RYDER:  Okay.  Your Honor, I'm marking this exhibit as exhibit --
Hearing Exhibit 12.

		JUDGE SOLOMON:  And without objection it is entered into evidence.

(Whereupon, the document referred to as Hearing Exhibit 12 was marked
and received in evidence.)

		JUDGE SOLOMON:  Okay, it's OSHA's turn.  Mr. Perry, are you --

		MR. PERRY:  Yes, Your Honor.

		JUDGE SOLOMON:  Okay.

		MR. PERRY:  Dr. Coble has a few questions to start us off.  And thank
you all for appearing today.  It's been very helpful.

		DR. COBLE:  Yes, hello.  And my first question would be addressing the
concern regarding the adoption of the ISO/CEN curve, and the potential
biases that could be introduced by that.

		The testimony mentioned a bias that could be as high as 30 percent,
but in the work of Soderholm where he examined 31 theoretical
distributions, that most of the biases were less than -- in 28 out of
the 31 it was less than 20 percent.

		So under what conditions and how realistic are they that a bias would
be as high as 30 percent?

		DR. HALL:  Now, you say, pretty categorically, if I can answer the
question for you, Joe, that the bias was baked in at 30 percent, is kind
of what his comment was, and that was a good thing, because it was more
protective.  I can get the paper for you and get the quote for you.  I'd
be happy to find it.

		DR. COBLE:  Yes, but you're aware that he evaluated 31 different
hypothetical distributions?

		DR. HALL:  Yes.

		DR. COBLE:  And that for 28 out of the 31 it was less than 30 percent?

		DR. HALL:  Yes.  I'm aware of that.

		DR. COBLE:  Right.  And for concentrations that are typically seen in
industrial environments, it was less than 20 percent.

		JUDGE SOLOMON:  I want to make sure that this is entered into the
record.  Mr. Hall, you're pretty far away from the microphone.  I want
to make sure that he got --

		DR. HALL:  Yes, I'm aware of that.

		DR. COBLE:  Okay.  Thank you.  I just wanted to clarify that.  The
next would be toward Dr. Knutson, and you indicated that you had done
some work at fracking sites.  Have you --

		DR. KNUTSON:  I'm sorry.  I missed that.

		DR. COBLE:  You indicated that you have been to several hydraulic
fracturing sites in your career, to observe.

		DR. KNUTSON:  Yes.

		DR. COBLE:  During your time at these sites, have you ever observed
dust controls that were in place?

		DR. KNUTSON:  I have seen dust control systems that were in place
during the site visits.  The -- I did not have data as to how well they
controlled.  You had a couple of incidences.  I had observational data,
and the observational data, for example, one had taken a sock and put it
on, somewhat similar to the NIOSH mini baghouse and put it on there.

		I don't think the media was the proper media.  It clearly had a
visible emission coming out of it, much reduced from what I would have
expected without it, but insufficient at that point.  

		So I've seen that.  I've seen dust collectors attached to the sand
movers themselves, trying to control the pneumatic conveying, attached
on the front end or rear end, depending on which way you want to look at
it -- the front end, from the perspective of the way that the flow is
moving, so it would be towards the dragon tail.

		And that was marginally acceptable, in that it did reduce some of the
plume that came out through the access ports, but probably not
adequately sized nor adequately designed for the application.

		DR. COBLE:  Have you ever been asked to provide recommendations on
dust control techniques at a fracking site?

		DR. KNUTSON:  I'm in the process of doing so, but I have not completed
that.

		DR. COBLE:  Do you have any awareness of the NIOSH industry
partnership, they call the Steps program, in which they're working
toward identification and implementation of dust controls?

		DR. KNUTSON:  I've read some documents on that, but I'm not totally
familiar with the program.

		DR. COBLE:  And then have you ever seen any exposure monitoring data
from hydraulic fracking sites, other than the data in the OSHA -- in
the NIOSH report that OSHA is utilizing?  Is there any additional data
that you're aware of that we could use for -- to enhance our analysis?

		DR. KNUTSON:  Yes, I have.

		DR. COBLE:  Would you be able to submit some of that for --

		DR. KNUTSON:  I've got confidentiality agreements, so I can't answer
that.  Just roughly speaking, I did not see a marked difference.

		DR. COBLE:  So that the NIOSH data may be fairly representative, in
your mind, of an uncontrolled situation at a fracking site?

		DR. KNUTSON:  Well, it's not completely erroneous.  It --

		DR. COBLE:  Yes, okay.

		DR. KNUTSON:  Obviously, the problem with many monitoring data that
occurs in hydraulic fracturing is that there's no two sites that are the
same.  You've got the situation where the number of hoppers that you've
got, the conveyors, the type of sand, the size of the sand, the length
of the campaigns, all of these things change.

		So there is considerable variability that occurs in any airborne
concentrations.

		DR. COBLE:  And your conclusion that it would not be feasible to
reduce exposures to below the proposed PEL, does that apply to all the
workers at the fracking site or just to a subset of those workers?

		DR. KNUTSON:  Well, obviously if you look at somebody that's working
in the lab, that their exposures are considerably different than the
exposures to people that you -- that are working down in the center
core area.

		DR. COBLE:  Right.  So some of the NIOSH data indicated that a
significant percentage of the workers at fracking sites are currently
exposed above the current PEL, and that implementation of controls could
bring 90 percent of the workers down to below the proposed PEL, so that
the only workers who would be above the proposed PEL are the sand
workers involved in the direct transfer of the sands.

		And the exposure to all the ancillary workers would be reduced to
below the proposed PEL.  Do you have an issue with that conclusion?

		DR. KNUTSON:  I have the same issue that I have with the feasibility
argument in the Appendix A.  I have not seen an argument that is
satisfactory to me that indicates that feasible engineering controls are
acceptable -- or have been found and are applicable and work in this
particular industry.

		DR. COBLE:  Okay.  Thank you.

		JUDGE SOLOMON:  Mr. Perry.

		MR. PERRY:  Next up will be Brian Albrecht from the Salt Lake City
Laboratory.  Just switch seats here, one second.

		MR. ALBRECHT:  Thank you.  My first comments -- or questions are for
Dr. Hall.  Dr. Hall, you mentioned in your testimony that there is a
range of geometric standard deviations that are likely to be encountered
in workplaces.

		DR. HALL:  Yes.

		MR. ALBRECHT:  And I was not able to write that number down.  Would
you repeat what it was?

		DR. HALL:  It was actually stated, I think, earlier by the NIOSH
panel, and it's what I've found in my experience, that typically the
range you see in an industrial setting -- in a controlled industrial
setting is somewhere in between about 1.8 to about 3, in that region.

		When you get up into the 3 or 4 region, that's a pretty uncontrolled
environment.

		MR. ALBRECHT:  So -- okay, 1.8 to 3, thank you.  According to the --
this is the ninth page of your written submission, titled -- hold on
just a moment here, "Comments on OSHA Proposed Rule: Occupational
Exposure to Respirable Crystalline Silica, Part 1."

		On the ninth page of that docket entry, you have the bias map for a
Dorr-Oliver cyclone at 1.7.  And it -- I'll describe the chart briefly
for the record.  Along the x-axis is the mass median diameter from 0 to
25, and along the y-axis is the geometric standard deviation from 1.5 to
3.5.

		If I start at the range -- I'm going to look at 2, because it's on
there and it's close to 1.8, but if I go from 1 -- sorry, if I go from
2 to 3, I'd like it noted in the record that the bias map predicts that
the bias is actually between 5 and 10 percent everywhere then, that you
are likely to encounter a aerosol in the workplace.

		It does not approach 30 percent until the geometric standard deviation
is much lower than that; is that correct?

		DR. HALL:  We're talking about two different things, I think.  The
bias that is -- that Soderholm discussed as built into the standard is
the difference between what would be collected under the convention that
was in play, under ACGIH and now the new convention that's in play,
which allows for a greater penetration of larger particles.

		MR. ALBRECHT:  So you're not --

		DR. HALL:  That bias is built in and not included when you're looking
at the results here.

		MR. ALBRECHT:  So you are conceding that there's not a large sampling
bias using the Dorr-Oliver at 1.7 L/min?

		DR. HALL:  I'm conceding that there is a difference in the two
conventions, and part of that is built in as -- as Dr. Coble was
talking about, there's a minimum of 20 percent but somewhere between 20
and 30 percent.  But Soderholm said that it was basically 30 percent
baked in and that was protective.

		MR. ALBRECHT:  All right.  You also mentioned that the GK 2.69, that
you had read a study that NIOSH reported that as they analyzed the
effectiveness of that sampler at a flow rate of 4.2 and what biases were
observed, in the same paper did you also read that the biases were
nonexistent at a flow rate of 4.4?

		DR. HALL:  Excuse me?  I didn't hear the last part of your question;
did I also read what?

		MR. ALBRECHT:  The biases that were observed at 4.2 flow rate in the
GK 2.69 pump, in the paper you've cited, in the same paper, NIOSH
observes that at a flow rate of 4.4, that bias is no longer important,
or no longer significant.  Did you also understand that from the paper?

		DR. HALL:  No, I did not.  And also, they've actually changed the flow
rate now to 4.8.  So there's quite a change of flow rates through this
particular sampler at which you can either experience or not experience
bias, and you wouldn't know that if you didn't really have a sampler to
show that.

		MR. ALBRECHT:  All right.  I have some comments -- or questions for
Mr. Lieckfield.  Mr. Lieckfield, you mentioned your laboratory, Bureau
Veritas, uses a 5 µg filter to determine the reporting limit, and that
it can range from 7.5 to below 5.

		MR. LIECKFIELD:  That's correct, as a pass/fail at the reporting
limit.

		MR. ALBRECHT:  How far below 5?  Would it also be 25 percent as 7.5
is --

		MR. LIECKFIELD:  Oh, it'd be much, much more than that, under 5.

		MR. ALBRECHT:  Okay.

		MR. LIECKFIELD:  Five is the -- you will see a deflection in the
x-ray method.  You will see a peak.  So we're looking at more of -- can
we see that.  And this discussion, just to make sure we're all
understanding this, the method can see down to 5 µg.  

		It's not a question of limit of detection or even limit of
quantitation.  It is the variability between that number and what is
being required, or would be required under the proposed rulemaking.

		MR. ALBRECHT:  But you also stated in your testimony as you answered
some of the questions from the participants that Veritas has not done
any studies to actually determine the reliable quantitation limit.  So
are you not sure if it is 5?

		MR. LIECKFIELD:  No.  We know it's 5 because we can see it, by a low
standard.  AIHA accreditation allows a reporting limit to be based on a
low standard that you have to see every time you set up a run of
samples.

		MR. ALBRECHT:  Yes, I agree with that.  But my understanding is that
AIHA accreditation requires you to be able to see that to plus or minus
25 percent.

		MR. LIECKFIELD:  That I am not familiar with, not at the reporting
limit.

		MR. ALBRECHT:  All right.

		MR. LIECKFIELD:  So you have information I don't have.

		MR. ALBRECHT:  But when you say that you haven't performed these
studies to determine the LOQ of the instrumentation that you use, if you
have not done that and you do not know for certain what the limit of
quantitation is, how is it in your testimony you can say certainly that
the proposed PEL is not feasible?

		MR. LIECKFIELD:  The proposed PEL has nothing to do with the limit of
quantitation.  I guess those are two independent issues.  So I may --
perhaps I'm not understanding your question.

		MR. ALBRECHT:  Can you not see any relation between the proposed PEL
and the limit of quantitation?

		MR. LIECKFIELD:  No.

		MR. ALBRECHT:  Okay.

		MR. LIECKFIELD:  As defined by the limit of quantitation?  How are you
defining the limit of quantitation?  I guess I asked a question after
all.

		UNIDENTIFIED SPEAKER:  Well, really.

		UNIDENTIFIED SPEAKER:  Okay.

		UNIDENTIFIED SPEAKER:  I would say nothing.

		MR. ALBRECHT:  Let me ask you a question.  You're recommending that
laboratories come to some kind of consensus about how a limit of
quantitation would be defined.  Do you have any recommendations for how
that could be accomplished?

		MR. LIECKFIELD:  We could use the NIOSH version of determining
LOD/LOQ, which we're familiar with, and that's running multiple samples
in and around the range of your predicted LOD and LOQ and doing a
statistical analysis on that.

		MR. ALBRECHT:  Is that the process that is described by Burkhart?

		MR. LIECKFIELD:  It is the process described by the NIOSH SOP on
LOD/LOQ.  I do not know the author of that.

		MR. ALBRECHT:  Very well.  Okay.  You also stated that OSHA's overall
inaccuracy was 26 percent without bias, and I'd like to know if you know
how that number was calculated.

		MR. LIECKFIELD:  Are you speaking about the OSHA --

		MR. ALBRECHT:  Yes.

		MR. LIECKFIELD:  -- method?  I'm citing what is in the published
method.

		MR. ALBRECHT:  I understand that.  Do you know how that number was
calculated?

		MR. LIECKFIELD:  No, I don't.  We'd have to ask OSHA that.

		MR. ALBRECHT:  Are you aware that it is --

		JUDGE SOLOMON:  I'll take administrative notice that he is OSHA.

		(Laughter.)

		MR. LIECKFIELD:  Well, then -- I'm --

		MR. ALBRECHT:  So --

		MR. LIECKFIELD:  Then I'm not sure why he's asking the question.

		MR. ALBRECHT:  Well, I asked the question because you said for sure
that that number was calculated without bias, but you --

		MR. LIECKFIELD:  No.  That's not a true statement.

		MR. ALBRECHT:  Oh, okay.  So you understand there is a bias in that
calculation?

		MR. LIECKFIELD:  According to OSHA, there is a 5 percent method bias. 
In the method, there is a 5 percent method bias, and a plus/minus 20
percent precision.

		MR. ALBRECHT:  Okay.  So you do understand how that -- you understand
the variables that went into that calculation --

		MR. LIECKFIELD:  Yes.

		MR. ALBRECHT:  -- as they are also listed in 142?

		MR. LIECKFIELD:  Yes, sir.

		MR. ALBRECHT:  Have you done any studies with your own data that
you've generated at Bureau Veritas to estimate, according to similar
calculations, the inaccuracy of the method that you use, on the
instrumentation that you use?

		MR. LIECKFIELD:  We have.  I don't have it with me, but it mimics what
both the OSHA method and the NIOSH method cites, in that same range.

		MR. ALBRECHT:  All right.  Fine.  Can you please submit that
information to the record?

		MR. CHAJET:  We'll consider it.

		MS. RYDER:  It sounds like information that OSHA would find really
useful in developing this --

		MR. CHAJET:  You provide the information we ask for, we'll provide the
information you ask for.

		UNIDENTIFIED SPEAKER:  That's not how it works.

		MR. CHAJET:  I know.

		JUDGE SOLOMON:  Again, for the record, these are two lawyers talking
to each other, really, and I'm not in position to order anybody to do
anything.  So you --

		UNIDENTIFIED SPEAKER:  Do you refuse to give this information?

		MR. LIECKFIELD:  Okay.

		JUDGE SOLOMON:  Just a minute.  I mean, if this actually is -- if you
already have the information, that's one thing.  If you have to come up
with the information by using some method of calculation, that might be
something else.  So I think the easy way to do this is, is the parties
talk to each other off the record.

		And then you -- if you have to, you can report back to me, and we'll
decide exactly how to do this.  Do you have any other questions?

		MR. ALBRECHT:  Yes.  

		JUDGE SOLOMON:  Reluctantly, you have other questions?

		(Laughter.)

		MR. ALBRECHT:  I have other questions.  Are you aware that OSHA no
longer uses the calculations that were used to arrive at 26 percent to
calculate the method inaccuracies of currently developed methods?

		MR. LIECKFIELD:  Assuming that's -- you must have changed it.  I am
taking everything off of the OSHA website and the OSHA methods.  So if
there has been a change that hasn't been published, then --

		MR. ALBRECHT:  No.

		MR. LIECKFIELD:  -- either I haven't -- then I must haven't -- I
must have not seen it.

		MR. ALBRECHT:  Well, I'm curious to know where you get the 25 percent
limit, upward limit that you've referred to several times, maybe even
collectively in your testimony.

		MR. LIECKFIELD:  That is in the -- it's in the benzene standard. 
It's in the sulfur dioxide standard.

		MR. ALBRECHT:  But have you read the current method development
guidelines that OSHA uses to develop methods?

		MR. LIECKFIELD:  I have.

		MR. ALBRECHT:  Did you see it in that document?

		MR. LIECKFIELD:  I don't recall.

		MR. ALBRECHT:  Okay.  The reason I ask is because -- and particularly
with your own data, I would be curious to know if you have looked at
recent OSHA biases, as far as PAT data, or even internally spiked
quality control, spiked media samples, or using your own data at Bureau
Veritas, if you would see similar inaccuracies reported, because at OSHA
we don't.  We find that no matter how the inaccuracy is calculated, it
is much better than reported on the face page of 142.

		I also want to make clear, do you feel like you have done enough work
to show that, analytically, the proposed PEL is not feasible?  Or are
you just saying that you don't think OSHA has done enough work to show
that it is feasible?

		DR. KNUTSON:  As I stated earlier, I --

		MR. ALBRECHT:  This question is directed to --

		DR. KNUTSON:  I think OSHA has not adequately demonstrated feasible
engineering controls.  I have not drawn a final conclusion myself.

		MR. ALBRECHT:  This question is not about engineering controls.

		DR. KNUTSON:  I'm sorry?

		MR. ALBRECHT:  This question is not about engineering controls.

		MS. RYDER:  It was directed toward Mr. Lieckfield.

		MR. ALBRECHT:  Analytically, the analytical method, has Bureau Veritas
done any research that proves that the proposed PEL is not feasible?  Or
is it your position that you just don't feel OSHA has done enough work
to show that it is?

		MR. LIECKFIELD:  I don't believe the industry has done enough to prove
method accuracy at the levels required.

		MR. ALBRECHT:  All right.  Are you aware that PAT data, in its
comparison, as it is used to compare inter-laboratory results, are you
aware that those results are never compared to the actual target that is
attempted to be spiked on a filter?

		MR. LIECKFIELD:  Well, actually it is.  The reference values must be
within 10 percent of the theoretical spiking in order for rounding
release.

		MR. ALBRECHT:  So they are compared to a number that is within 10
percent of the target --

		MR. LIECKFIELD:  That's correct.

		MR. ALBRECHT:  -- but not the theoretical target?

		MR. LIECKFIELD:  Not exactly, because -- well, you wouldn't run a PT
round.  You use reference labs to establish the average.

		MR. ALBRECHT:  Understood.

		MR. LIECKFIELD:  You will always get some differences in spiking.

		MR. ALBRECHT:  Okay.  I also want to talk just briefly about, you
made -- well, this is a question.  I keep saying comment, but these are
questions. 

		You stated that between 40 and 50 µg of analysis, that the error is
higher than it is between 100 and 200 µg.  So in other words you, in
your testimony, have explained that the lower microgram loading on a
filter will necessarily equate to more inaccuracy?

		MR. LIECKFIELD:  That's correct.

		MR. ALBRECHT:  Have you reduced any PAT data or internally spiked data
from Bureau Veritas that demonstrates that phenomenon?

		MR. LIECKFIELD:  I think the PAT data does demonstrate that.  That was
what I was citing.  There is no data below 40, in terms of PAT samples.

		MR. ALBRECHT:  Well, that may be true, and that leads me to a
different question.  Is there enough low spiked PAT data to be of
statistical significance to say really anything concrete about analysis
at that level, using PAT data?

		MR. LIECKFIELD:  No, there is not.

		MR. ALBRECHT:  All right.

		MR. LIECKFIELD:  There was five sets.  I think that is actually -- to
make sure that my testimony is clear, that's my whole point, is there is
not enough data to characterize it.

		But based on the data we do have, and general analytical imprecision,
the lower you get when you start pushing the instrument limitations, it
should expand.  So if you're at 30 percent at 50, you could surmise that
it's going to be greater than 30 percent below 50.

		MR. ALBRECHT:  So you say it should expand?

		MR. LIECKFIELD:  It should.

		MR. ALBRECHT:  But would you also be interested or surprised to know
that I have done statistical calculations like that with internal OSHA
data, and the lower spike samples actually have better accuracy and
precision than the sample set as a whole?

		MR. CHAJET:  I'm going to object and move to strike, and ask for any
data that they've done internally to be made public and part of the
record, and to ask the witness to stop being argumentative and
presenting his own testimony and instead just ask a question.

		This was not tolerated when the tables were turned and we were allowed
three questions of OSHA, and this is getting out of control.  So I ask
the Judge to please control the witness.

		MR. MOAR:  Your Honor?

		MR. CHAJET:  And have him make available for the record whatever
internal OSHA study he's referring to.

		MR. MOAR:  Your Honor, I would like --

		JUDGE SOLOMON:  Yes?

		MR. MOAR:  -- to point out that under the procedures --

		JUDGE SOLOMON:  Are we -- are you on a microphone, so that we get
your statements?

		MR. MOAR:  Sorry.  It's on?  I just want to point out that under the
procedures for this hearing, striking evidence is prohibited.

		JUDGE SOLOMON:  Well, there is a question whether it's actually
evidence.  I mean, he's not really testifying.  He's asking a question. 
So this is an example of what we had before, which is the loaded
question.

		And at this point, if the witness can answer, so be it.  So can you
respond to that?

		MR. LIECKFIELD:  The question is, have I seen OSHA's data showing
better precision at lower levels?  No.  If that was the question.

		JUDGE SOLOMON:  Now, this -- the matter about asking to produce
evidence is another matter altogether, and really this is not an
evidentiary hearing.  It's rulemaking.  And there are procedures to get
that information.  And among them, I guess you've already done, which is
a FOIA, to get this information; is that right?

		MR. CHAJET:  We will certainly do one for this now.

		JUDGE SOLOMON:  Well, you've already done them, from what I
understand.  So those are pending.  And --

		MR. PERRY:  Your Honor, we --

		JUDGE SOLOMON:  -- so I think we can move on to the next question.

		MS. RYDER:  One minor point.

		MR. PERRY:  I will make the point.  We will make the data from Salt
Lake City Lab that was referred to, we'll put that in the record.  I
don't know how long it will be.  It will certainly be in -- before the
end of --

		JUDGE SOLOMON:  Well, we have three weeks here --

		MR. PERRY:  -- post-hearing, so --

		JUDGE SOLOMON:  -- so you have additional witnesses that you could
use to put that on, and you can reduce it to writing and then we can
actually make it an exhibit and enter it into evidence.

		MS. RYDER:  I think it'll be probably part of the post-hearing comment
period.

		MR. PERRY:  I think that's probably more likely.  It just has to go
through a bit more --

		JUDGE SOLOMON:  We haven't gotten to that bridge yet, that -- we had
a discussion about that already.

		MR. PERRY:  Okay.

		JUDGE SOLOMON:  So I'm going to hold in abeyance anything about that. 
But it could very well be that you have the capacity within the next
several days to have some kind of document you can pass among
yourselves, and then we'll decide how to handle it.  

		Okay, do you have another question?

		MR. PERRY:  I think David O'Connor has a few questions for the panel.

		MR. O'CONNOR:  Yes, thank you. 

		Dr. Bunn, I just had a point of clarification in light of the
questioning from Dr. Sivin.  In your testimony, you had indicated that
over your 30-year career overseeing about 100,000 workers, you never saw
a case of silica-related disease.  Do I understand that to be limited to
silicosis cases?

		DR. BUNN:  Yes.

		MR. O'CONNOR:  Okay, thank you.  

		Dr. Knutson, you have some very extensive experience in designing
ventilation systems to control silica, I understand; is that correct?

		DR. KNUTSON:  Yes.

		MR. O'CONNOR:  In your work, do you rely on industrial hygiene
sampling results to help inform your designs?

		DR. KNUTSON:  When it's available.  It is a useful mechanism to
evaluate the situation as we start to look at what has to be done to
control.

		MR. O'CONNOR:  Is that type of information typically available?

		DR. KNUTSON:  Publicly available?

		MR. O'CONNOR:  Typically.

		DR. KNUTSON:  Oh, typically available?  We -- I don't know if there's
such a thing as a typical client for us.  We have several clients that
are quite small that have almost no ventilation.  We have clients that
are large and sophisticated that have their own industrial hygiene
departments.  And there the data is often available and extensive.

		MR. O'CONNOR:  Okay.  When you do use that data, do you know what type
of cyclones were used to obtain the data?

		DR. KNUTSON:  Sometimes I do, sometimes I don't.

		MR. O'CONNOR:  Of the times where you have that knowledge, do you
recall what types were used?  The Dorr-Oliver, Higgins-Dewell, SKC,
others?

		DR. KNUTSON:  I think it's been mostly the 10 mm nylon cyclone.

		MR. O'CONNOR:  Okay.

		DR. KNUTSON:  At 1.7 L/min.

		MR. O'CONNOR:  From your experience, is the type of cyclone that's
used for sampling typically a consideration in the development of the
industrial ventilation system?

		DR. KNUTSON:  Not as extensively as an understanding of the
person-machine interface.

		MR. O'CONNOR:  Does that imply that it is used in some respects?

		DR. KNUTSON:  I don't -- I didn't follow.  I'm sorry.

		MR. O'CONNOR:  What I'm really getting at is if you have a different
type of cyclone that's used, if you have information indicating that the
Dorr-Oliver or the Higgins-Dewell or the SKC cyclone was used to obtain
the sampling results that you're relying on, does that make any
difference in terms of the design of the ventilation system?

		DR. KNUTSON:  Probably not meaningfully.

		MR. O'CONNOR:  Okay, thank you.  Dr. Hall, you have also done some
assessment of worker exposures to silica.  What types of cyclones have
you used?

		DR. HALL:  Dorr-Oliver, Higgins-Dewell, and SKC.

		MR. O'CONNOR:  And when you're deciding what type of cyclone to use,
how do you make that decision?  What factors come into play there?

		DR. HALL:  Typically what was available.

		JUDGE SOLOMON:  Let's make sure you have a microphone.

		DR. HALL:  I'm sorry.  Typically what was available at the time, or
what we had rented from the supplier.

		MR. O'CONNOR:  Okay.  So you're not generally making some distinction
between the different cyclones, in terms of what's going to be
appropriate for a particular sampling situation?

		DR. HALL:  Not historically.  I think after review of the cyclones,
I'd probably make different decisions today.

		MR. O'CONNOR:  All right.  Thank you very much.

		JUDGE SOLOMON:  Mr. Perry?

		MR. PERRY:  I have just a couple of questions.  If I could just follow
up, Dr. Hall, on what you just said, what kind of decision would you
make today, knowing what you know?

		DR. HALL:  Knowing what I know, I would actually collect samples to
get an idea of what the aerosol distribution was like, what the particle
size distribution was like before I chose a cyclone and a flow rate.

		MR. PERRY:  Okay.  I need to ask a question that I actually asked this
morning, regarding Dr. Borak's testimony, but he said I should ask you
all, so I'm going to do that.

		In his testimony he recommended that OSHA define respirable
crystalline silica in our proposed rule differently than we have, in
part by saying that it represents or it's comprised of all the particles
that are less than 10 microns in aerodynamic diameter.

		If we were to do that, and then express an exposure limit in terms of
a gravimetric measure or some gravimetric measure of respirable
crystalline silica, is there any device or approach you can think of
that could measure respirable crystalline silica, all the particles that
are less than 10 microns in diameter?

		MR. CHAJET:  Objection, Your Honor.  Just to make sure the record is
clear, that's not what Dr. Borak said.

		MR. PERRY:  Well, that's --

		MR. CHAJET:  Maybe that's what you wrote down, but it's not what he
said.

		MR. PERRY:  Let me see.

		JUDGE SOLOMON:  Well, the easy way is to read back the transcript of
what he said, but --

		MR. PERRY:  Well, I have his written testimony right here, if I may.

		JUDGE SOLOMON:  Okay.  You can refer to the written testimony.

		MR. PERRY:  It's on Page 19 of his testimony, he says, "Second, the
definition should state that RCS is defined by particle size, e.g.,
particulate less than or equal to 10 microns."				JUDGE SOLOMON:  Okay,
so wait a second.  So do you remember that as being in your testimony?

		MR. PERRY:  It's Dr. Borak's --

		UNIDENTIFIED SPEAKER:  He's not here.  This is Dr. Borak's testimony.

		JUDGE SOLOMON:  Oh, I'm sorry.

		MR. PERRY:  I had asked him about this, and he thought the afternoon
panel would be a better group to ask.

		JUDGE SOLOMON:  Actually, we should have brought Dr. Borak back, but
that's --

		MR. PERRY:  Oh, I'm just --

		JUDGE SOLOMON:  Okay.

		MR. PERRY:  -- doing what he suggested I do this morning.

		MR. CHAJET:  You need to direct that question to Dr. Borak.  You
know --

		MR. PERRY:  I did this morning.

		MR. CHAJET:  You know as well as I do that the deposition curve for
particle size under the ACGIH is X percent at 10, Y percent at 9 --

		MR. PERRY:  Yes.

		MR. CHAJET:  -- Z percent -- right, so it's much more likely he's
referring to that, and saying all particles below 10 should be added up,
counted and dumped in the box.

		MR. PERRY:  Okay, but --

		JUDGE SOLOMON:  Well --

		MR. CHAJET:  You know, and you misstated --

		JUDGE SOLOMON:  Just a second.  So the Solicitors -- who's going to
hand this?  Get to refer to the --

		MS. RYDER:  Maybe if Mr. Perry wants to continue asking the question?

		MR. PERRY:  I'm happy to move on.  I'll just say that I did ask him
this question this morning, and he thought I should ask this group.  I
did read his exact words from his own statement, so --

		JUDGE SOLOMON:  There is a way that you could just ask him if he's
familiar with the testimony, that --

		MR. PERRY:  Okay.

		JUDGE SOLOMON:  -- that was introduced this morning and read the --
and you've already read that into the record.

		MR. PERRY:  Okay.

		JUDGE SOLOMON:  So the question is, are you familiar with that?

		MR. PERRY:  Are you familiar with Dr. Borak's recommendation in his
testimony how OSHA should define respirable crystalline silica in its
standard?

		JUDGE SOLOMON:  And the answer is?

		DR. HALL:  Not really, no.

		JUDGE SOLOMON:  Okay.

		MR. PERRY:  Okay, very well.  Dr. Hall, you had mentioned, I
think -- I want to make sure I heard this and got this down correctly. 
You said today that pump error is minor?

		DR. HALL:  Pump error is a minor contribution --

		MR. PERRY:  And that --

		DR. HALL:  -- in overall error.  What?

		MR. PERRY:  -- differences in flow rate are not significant?

		DR. HALL:  Typically are not significant.  They -- even the
fluctuations that I've been observing, basically studied by NIOSH,
didn't show a great impact on the samples that were collected, on their
distribution or their mass.

		MR. PERRY:  Is that -- what you're referring to, is that -- are you
are referring to pump performance over full shift sampling?

		DR. HALL:  As I recall, yes.  Lee did basically full shift sampling,
yes.

		MR. PERRY:  Okay, very good.  Thank you.  That's all I had.  So I
think Ms. Ryder has some questions.

		MS. RYDER:  Sure, I have a few questions.  Mr. Lieckfield, I think
I'll start with you.  Can you tell us a little bit about what type of
sampler or samplers and analytical methods you're currently using in
your laboratory?

		MR. LIECKFIELD:  Analytical methods, x-ray diffraction.

		MS. RYDER:  And what -- do you know what type of samplers are being
used?

		MR. LIECKFIELD:  No.

		MS. RYDER:  Okay.

		MR. LIECKFIELD:  We have -- as a commercial laboratory, we do not
have knowledge of all the various sampling apparatus.

		MS. RYDER:  I guess maybe I should say what type of samplers do you
have the capability -- filters from which samplers do you have the
capability of analyzing?  Would it be anything that would be analyzed
with x-ray diffraction, or infrared methods?

		MR. LIECKFIELD:  Let me -- I believe --

		MS. RYDER:  Go ahead.

		MR. LIECKFIELD:  -- this might answer it.  The --

		UNIDENTIFIED SPEAKER:  Your Honor, would you have him please use the
microphone?

		JUDGE SOLOMON:  Yes, I'm sorry.

		MR. LIECKFIELD:  Sorry about that.  The typical sample we get in, it
would be a 37 mm PVC filter, 25 mm PVC and silver membrane.  That's the
actual collection media, as opposed to the sampler.  Be it a Dorr-Oliver
and that, we don't have knowledge of what our clients use.  Does that
answer the question?

		MS. RYDER:  Yes.  That's what I was --

		MR. LIECKFIELD:  Okay.

		MS. RYDER:  -- wondering.  And can you tell us a little bit about how
you are reporting respirable crystalline silica lab results back to your
customers?  Are you telling them the results and then saying what the
sampling and/or analytical error associated with that result would be?

		MR. LIECKFIELD:  Generally speaking, we're reporting a mass and an
airborne concentration, providing the client gives us air volume
information, so mass and milligram per cubic meter.

		We also, at -- given the current OSHA standard, we also provide a
percent, based on the particulate mass.  So we're doing both a
gravimetric analysis and then follow it by silica on the filter.

		MS. RYDER:  And are you including any error with that reporting?

		MR. LIECKFIELD:  For some clients, we do.

		MS. RYDER:  What is --

		MR. LIECKFIELD:  Most, we don't.

		MS. RYDER:  Okay.  What is that error?  How are you determining what
that error is?

		MR. LIECKFIELD:  We determine that error through lab control samples,
which are spiked at 50 µg.

		MS. RYDER:  And are these your own -- this is something that you've
come up with, within your own laboratory?  Or is this something based on
error associated with a particular method or the x-ray diffraction?

		MR. LIECKFIELD:  The lab -- standard practice quality control in the
lab industry would be to prepare known spiked samples, typically called
laboratory control samples.  And that is in a duplicate.

		Those are run through the analytical process, with a set of client
samples.  That -- it tests the quality of the sample preparation as
well as the analytical calibration.

		MS. RYDER:  And what kind of numbers are you giving for that error?

		MR. LIECKFIELD:  In the -- what we're using at the 50 µg is in the
range of plus/minus 20, 25 percent.

		MS. RYDER:  Okay.  How would that compare to other substances that you
might analyze, like asbestos?

		MR. LIECKFIELD:  Asbestos is not a good example.

		UNIDENTIFIED SPEAKER:  No.

		MR. LIECKFIELD:  Because that's actually worse.

		MS. RYDER:  Well, I mean, can you tell me what asbestos would be?

		MR. LIECKFIELD:  No.  I don't know off the top of my head.  You're
talking about fiber counts error?

		MS. RYDER:  Basically, right.

		MR. LIECKFIELD:  Asbestos fiber counts?

		MS. RYDER:  Right, right.

		MR. LIECKFIELD:  I would hazard a -- just a somewhat educated guess
of plus/minus 50 percent at the low end.  The asbestos fiber count
method is just known to not be very precise.

		MS. RYDER:  Okay.

		MR. LIECKFIELD:  Now, if you look at organic solvents and metals, the
overall error at the low end of mass loadings would be in the range of
plus/minus 15 percent.

		MS. RYDER:  All right.  I think I have a couple more questions for
you.  Can you tell us a little bit about the way that your lab works to
minimize variation in sample analysis, using XRD, like how you would
account for potential interferences?

		MR. LIECKFIELD:  For -- well, the lab control samples, as I said,
measure if there is any error in the process, and it's controlled
through -- the overall error is controlled through training,
calibration, qualification of the analysts.

		I just lost my train of thought.

		MS. RYDER:  I guess I'm thinking for -- to account for any potential
interferences, are you --

		MR. LIECKFIELD:  Right, I'm sorry.  Yes.

		MS. RYDER:  -- doing any acid washing or anything like that?

		MR. LIECKFIELD:  Yes.  I'm sorry.

		MS. RYDER:  Okay.  No problem.

		MR. LIECKFIELD:  I apologize.  For interferences, we look at the
primary diffraction angle, then the secondary.  And we try to get --
fit a ratio between the two, which for us is about 1 to 5.

		MS. RYDER:  I have a more general question about how your lab works. 
What is the, I guess, standard turnaround time that you have from when
you get a sampling and when you're able to give results back to your
customers?

		MR. LIECKFIELD:  Standard turnaround is five days cited, probably
seven days actual -- business days.

		MS. RYDER:  Okay.  I guess -- said another way, what's your
capability of getting these results back?

		MR. LIECKFIELD:  Capability is -- you're talking about the shortest
possible time?

		MS. RYDER:  Right, I guess so.

		MR. LIECKFIELD:  Twenty-four hours.

		MS. RYDER:  Okay.  All right, thank you.

		MR. LIECKFIELD:  With notice, though.

		MS. RYDER:  With notice?  That's fair.

		MR. LIECKFIELD:  In case there's any -- and there's some money
involved, too.

		MS. RYDER:  Of course.

		MR. LIECKFIELD:  There is a gentleman over in --

		(Laughter.)

		MS. RYDER:  Okay, Dr. Hall?  I have a few questions for you.

		DR. HALL:  Yes.

		MS. RYDER:  I just wanted to follow up on your critique of the NIOSH
study of the hydraulic fracturing sites, to see if you know of any other
studies that aren't in the record that would be more representative of
the hydraulic fracturing industry and types of exposures that they're
seeing there.

		DR. HALL:  I know of other studies.  I don't have access to them, and
they are being held in confidence by the people who conducted them.

		MS. RYDER:  So these aren't -- this is not anything published?

		DR. HALL:  Not public-dom, but --

		MS. RYDER:  Okay.

		DR. HALL:  -- the industrial hygiene community has been fairly active
in assessing exposures from the fracking industry.

		MS. RYDER:  Okay.  And I just have a question about a couple of
studies that I think you were involved and you conducted.  One is the
assessment of exposure to airborne silica in foundry environments, and
the other is crystalline silica exposures in coal and non-coalmining
environments.  Do you know if those have been submitted to the record
already?

		DR. HALL:  Not that I am aware of, no.  I didn't rely on them for
reference material, so I didn't submit them, no.

		MS. RYDER:  Do you think that's something that could be added to the
record for OSHA to consider?

		MR. CHAJET:  We'll consider it.

		MS. RYDER:  Okay.  All right.  Do you think you could give a quick
overview of the types of local exhaust systems that you designed and
controlled to control exposure to silica in foundries?

		DR. HALL:  Are you talking to me about that?

		MS. RYDER:  Yes.

		DR. HALL:  And not Gerhard?  Typically --

		MS. RYDER:  We can later.

		DR. HALL:  -- they were AIHA designs that we absconded from the
ventilation handbook and had constructed and installed.

		MS. RYDER:  Do you know how well those were able to control exposures
among workers?

		DR. HALL:  In the environments that we put them in, they were able to
control exposures fairly well.  I say fairly well.  If you don't keep up
maintenance on them and keep an eye on them, they get turned around
after a short period of time.  The industry is fairly rough on
equipment.

		So, many of them that I know that I helped install, you go back a
couple of months later and they would be degraded, and you have to
basically update them.

		MS. RYDER:  All right.  Thanks.  

		Dr. Bunn, I have a few questions for you, if that's okay.

		JUDGE SOLOMON:  Yes, shift -- let's shift the microphone.

		MS. RYDER:  A lot of -- I think you've talked a lot about your
experience as a medical director, and I was just wondering if you can
talk a little bit more about your tenure at maybe the first place that
you discussed, where you worked, where it was a company that was
involved in diatomaceous earth mining.  Can you tell us about the
medical surveillance program that you -- I think you said that you had
there and, you know, how -- which workers participated in that?

		DR. BUNN:  Yes.  First I should explain that that site's been under
study for many, many years.  So the answer is, they receive yearly
examinations, yearly PFTs, chest x-rays, you know, because of the
radiation, we varied a bit. 

		Everybody received exposure measurements in all of the areas they
were -- anyone who wore respiratory protection went through the
appropriate respiratory protection programs.  So that's what we did.

		Now, I should explain to you that we would not necessarily do that in
every case.  Fiberglass workers, we did every year, but we didn't look
at, for example, petrochemical workers, other workers, forest products. 
But for that particular population, that's what we did.

		MS. RYDER:  This is in the same plant, or same facility, as opposed to
the same company?

		DR. BUNN:  Yes.  That's Lompoc.  That's the diatomaceous earth.

		MS. RYDER:  Okay.  And what -- did you say those were annual medical
exams?

		DR. BUNN:  Yes.  Generally, they were.

		MS. RYDER:  Okay.  And I think you might have said it earlier, but did
those continue into retirement, and if so, how long?

		DR. BUNN:  The answer to that is, in that particular case some of them
did.  But, again, you got to -- I mean, it's a research setting.  It
generally would not continue into retirement.

		The question you brought up, that's brought up earlier, which is did
we leave control?  The answer is in a generally and, you know, for lots
of support, I think, generally unionized group, for example, at
Navistar, and we did pretty much the same thing in Manville.  We follow
guys until their death.

		Now, we don't get routine exams, but we do get health information.  We
have health promotion programs.  We monitor people who have
cancer-related disease, who have pulmonary-related disease.  They're all
in a database that we track regularly.

		We look at healthcare, we look at healthcare costs.  We look at
different interventions that we've been looking at to improve health,
and that continues.  Our biggest cost in our current company is about 3
to 1 retirees.  So we don't forget about the guys, for sure.

		MS. RYDER:  You were saying not all of them got those medical exams
into retirement?

		DR. BUNN:  Well, they were eligible for a yearly exam, yes.  We don't
know --

		MS. RYDER:  And do you know how many actually did?  I think this might
have been a question asked earlier, but --

		DR. BUNN:  Yes.  You know, the answer is, we don't really know.  If
you look at a general population, you may only get 10 percent.  But if
you get people that are exposed, and then we have very aggressive
contacts from both management and union, so I don't know the number, but
it was significantly higher.

		MS. RYDER:  Okay.  And can you tell us a little bit about the types of
exposure that those workers had?  Do you know at what level they were
exposed?

		DR. BUNN:  They were highly variable.  If you look at the PEL, we
exceeded the PEL in a number of the -- I mean, the way the mine worked,
the mine was crystalline silica.  It was mostly diatomaceous earth,
which is a -- when we went to milling, went to calcining.  We came out
with cristobalite, which is more like 60, 70 percent crystalline silica.

		The highest exposure levels to those are like baggers, people who put
things on there.  And their exposure levels were significant.  They were
well above 100.

		In the foundries, even with all of our efforts, they were commonly
over 100.  I mean, we -- one foundry, I'll be honest, kind of 50/50. 
One foundry was always over 100.  We just -- there was just no way --
in several processes.

		I won't say, not the whole plant, but in several processes there was
no way to get to that, and that's true today.

		MS. RYDER:  And this may have been something you already answered so
forgive me, but were you -- in all of these different companies where
you worked, were you -- and you were a medical director, were you
looking at both silicosis, or also other silica-related diseases?

		MR. BUNN:  The -- my testimony was related to silicosis, but of
course we looked other related diseases, would be of a hard call, you
know.  We looked at COPD.  We had programs for that.  But, you know, we
can argue -- we had 70 percent smokers, so it's very hard to -- yes,
we looked at lung cancer.  Yes, we looked at COPD.

		Was it -- we had lots of other reasons, so that was a very complex
issue.  But I was speaking specifically with silicosis, because it gets
so confounded.

		MS. RYDER:  I guess, then did you see those other diseases among the
workers?

		DR. BUNN:  Pretty much as you would expect.  I -- you know, I
mentioned DEERS.  We did a -- with the UAW, we did a joint study, and
it gave pretty much -- what we came out with was pretty much exactly
what you would expect, a little less, given the number of smokers.  It's
not a very specific study, by the way, but not published.

		MS. RYDER:  Okay.  All right, I think that is all of my questions for
you, Dr. Bunn.

		DR. BUNN:  Thank you.

		JUDGE SOLOMON:  Mr. Perry, anything else?

		MR. PERRY:  If we could have one question as a follow-up, Daniel
Johansen has a question for Dr. Lieckfield.

		JUDGE SOLOMON:  Sure.

		MR. JOHANSEN:  You talked about earlier --

		JUDGE SOLOMON:  Actually, we're ahead of the game here, so --

		MR. PERRY:  Okay.  Well, maybe there'll be two questions.

		MS. RYDER:  Maybe we have more.

		JUDGE SOLOMON:  Okay.

		(Laughter.)

		MR. JOHANSEN:  Okay, since I'm given freedom for more -- but you
talked about reporting off of the primary peak on your analytical
samples, and then going to your secondary peak for a qualitative
confirmation; is that correct?

		MR. LIECKFIELD:  That's correct.

		MR. JOHANSEN:  If you have a mineral matrix that interferes with these
two peaks, what is your procedure in your laboratory?

		MR. LIECKFIELD:  If there is a sufficient mass of quartz, we would now
then look at the tertiary peak --

		MR. JOHANSEN:  Okay.

		MR. LIECKFIELD:  -- to try to resolve.  And depending on what the
interference might be, there is some chemical treatments.  But where --
our practice is to do all of those chemical treatments right from the
start.

		MR. JOHANSEN:  Okay.  Do you ever look at the quaternary peak?

		MR. LIECKFIELD:  No.  We cannot see that using our procedure.

		MR. JOHANSEN:  You talked about your lab control samples being spiked
at 50 µg?

		MR. LIECKFIELD:  Correct.

		MR. JOHANSEN:  Is that correct?  And that's -- is there a time when
those do not pass what -- where they'd fall out of control?

		MR. LIECKFIELD:  I'm sure there is times that they're out of control,
just statistically.  We use two standard deviations, so we have a 5
percent probability something would be out.

		If they do go out on a particular batch of samples, we do an
investigation as to what would be the cause.  If we can determine the
cause, we'll then make a decision on whether that affects the field
samples.

		If the cause is related to the field samples, we make that very
difficult call to the client and tell them we've just voided all their
samples.  That doesn't happen much, just for the record.

		(Laughter.)

		MR. LIECKFIELD:  It happens from time to time.  If in the same
investigation, if we believe that through our fact-finding that the
outlier was just -- you know, because of the quality control samples
and did not affect the field samples, then we'll report the field
samples with a note that describes the quality control.  Does that --

		MR. JOHANSEN:  Earlier you mentioned that you do not determine the
particle size distribution on the samples that you receive from your
lab; is that correct?

		MR. LIECKFIELD:  That's correct.

		MR. JOHANSEN:  How do you know if there are large particle sizes that
are biasing your results?

		MR. LIECKFIELD:  We don't.

		MR. JOHANSEN:  Okay.  I think that's all the questions I have.

		JUDGE SOLOMON:  Any other questions, Mr. Perry?

		MR. PERRY:  My panel keeps thinking of additional questions, so I
think Ms. Ryder has a couple more.

		MS. RYDER:  I have, I think, two questions for Dr. Knutson.  You
mentioned that you're engaged in hydraulic fracturing projects
concerning silica control, and I was just wondering what methods you're
currently using or recommending to your clients?

		DR. KNUTSON:  Could you repeat the question?  I'm hard of hearing. 
One of the advantages of being old is that I'm old.  One of the
disadvantages is that I have a hard time hearing.

		MS. RYDER:  I'll speak up a little bit.  You mentioned that you're
engaged in hydraulic fracturing projects concerning silica control, and
I'm just wondering what types of methods you're using or recommending to
companies to implement.

		DR. KNUTSON:  It's difficult to answer that because I'm in process and
have not finalized my conclusions.  It's also difficult to answer that
because I'm under confidentiality agreements.

		So the answer that I'm going to try to give is vague, and I'm really
trying to avoid answering it, but I will try anyways.  Certainly, we are
looking at local exhaust ventilation as a major tool.  We're looking at
material handling as a major tool.

		Is that -- both of those approaches have to be used in order to be
able to successfully reduce -- substantially reduce the potential dust
generation.  And so they kind of work together.

		And then we get back to the third part of it, which is observing and
understanding the person-machine interface, and where the operator is
and what the operator has to do in order to be able to get all of that
to come together.

		So that's more Ventilation 101 than answering your question, but
because of confidentiality issues, it's difficult to be more specific.

		MS. RYDER:  I guess maybe you can say that you do have clients that
you're developing these ventilation or different controls for.  It's not
like you're telling them I can't help you.

		DR. KNUTSON:  I'm in the process of developing.  I have not completed
that yet.

		MS. RYDER:  Okay.  So you're not turning away anyone, at this point,
because it can't be done?

		DR. KNUTSON:  Oh, at this point, no.  I'm a consultant.  There's only
one answer I've got.  It's yes.

		(Laughter.)

		MS. RYDER:  It can be done, okay.  Okay.  To follow up on something I
think Mr. O'Connor asked you about, in how you make the determination
that a control is working.  Sometimes -- I think you said you have
sampling information.

		DR. KNUTSON:  One of the most frustrating aspects of being a
consultant is that when the project is over, it's done.  And that means
they turn it on, and now they no longer are concerned about it, and I'm
not in the loop.

		And it's frustrating to have systems that I have developed, designed,
and have installed and trouble-shot at the beginning to get them going,
but then no longer am employed or retained by the client to follow it
all the way up.	

		So in a lot of cases, it's looking at the original data that was
collected, where the volumetric flows were expected, the static
pressures, the fan operating correctly, the baghouse operating
correctly, those kinds of things I have information on.  What the
exposure levels are, I don't often have that kind of data.  It's
frustrating.

		MS. RYDER:  Okay.  I think that's everything.  Thank you all very
much.

		MR. CHAJET:  Thank you.  I move to close the record for the day, Your
Honor.

		MR. PERRY:  I think we are actually done now, Your Honor.  Thank you.

		JUDGE SOLOMON:  Okay.  So that's unanimous.  So we'll start again
tomorrow at 9:30.  Is there anything else that anybody wants to bring up
before we go off the record?

		MS. RYDER:  I guess one thing; did any of you have written remarks
that I could include as hearing exhibits?

		MR. KNUTSON:  The only written remarks I have is a report that I
believe has been submitted to the record.

		MS. RYDER:  Okay.  Okay.

		MR. PERRY:  We do have -- I have the picture that they're going to
give, that was actually part of a document that they've proffered, that
was, I guess, part of the slide presentation.  Is that right?

		MS. RYDER:  Oh, this is going to be part of the same slide that --

		MR. PERRY:  So we have -- I haven't admitted it into evidence --

		MS. RYDER:  Okay.

		MR. PERRY:  -- but that's the same picture.

		MS. RYDER:  You don't need to.

		MR. PERRY:  It's the same picture.

		MS. RYDER:  Okay.  I think that's going to be e-mailed.

		JUDGE SOLOMON:  Well, you're going to talk to each other about --

		MS. RYDER:  We don't need to.

		JUDGE SOLOMON:  Well, you have other things to talk to each other
about.  Okay, so if there's nothing else, the hearing is closed.

		(Whereupon, at 4:44 p.m., the hearing was continued, to resume the
next day, Thursday, March 20, 2014, at 9:30 a.m.)

C E R T I F I C A T E

	This is to certify that the attached proceedings in the matter of:

INFORMAL PUBLIC HEARINGS FOR THE PROPOSED RULE 

ON OCCUPATIONAL EXPOSURE TO

RESPIRABLE CRYSTALLINE SILICA

March 19, 2014

Washington, D.C. 

were held as herein appears, and that this is the original transcription
thereof for the files of the United States Department of Labor,
Occupational Safety & Health Administration.

			      						         					____________________________

			         	ED SCHWEITZER

			         	Official Reporter

		

_________________________

		Continued

  PAGE  356 

Free State Reporting, Inc.

1378 Cape St. Claire Road

Annapolis, MD 21409

(410) 974-0947

			   PAGE  550 

